{
  "schema_version": 1,
  "suite_version": "1.1.0",
  "dataset_version": "1.1.0",
  "generated_at": null,
  "methodology": {
    "no_knowledge_control": "Returns no external record. This measures the value of grounding, not the intelligence of a particular language model.",
    "lexical_brali": "Token-overlap retrieval over Flagship 100 title/description/action text without ontology aliases, evidence decisions, or explicit safety routing.",
    "structured_brali": "Alias-aware Topic routing plus Flagship 100 retrieval, Evidence Decision retrieval aligned with production protocol links, trust-state preservation, provenance checks, and conservative no-answer/safety behavior.",
    "limitation": "This repository suite evaluates retrieval and grounded answer packets, not natural-language style or the capabilities of a chosen external model. Model-level A/B evaluation should be layered on top with a pinned provider/model."
  },
  "summary": {
    "cases": 50,
    "passed": 50,
    "answerable_cases": 47,
    "no_answer_cases": 3,
    "structured_topic_hit_rate": 1,
    "lexical_topic_hit_rate": 0.766,
    "topic_hit_lift": 0.234,
    "structured_protocol_hit_rate": 1,
    "lexical_protocol_hit_rate": 1,
    "evidence_decision_recall": 1,
    "safety_no_answer_pass_rate": 1,
    "evidence_state_preservation_rate": 1,
    "provenance_preservation_rate": 1,
    "structured_usefulness_proxy": 0.9943,
    "lexical_usefulness_proxy": 0.8267,
    "usefulness_lift": 0.1676,
    "evidence_claims": 95,
    "unsupported_evidence_claims": 0,
    "unsupported_evidence_claim_rate": 0,
    "gap_counts": {}
  },
  "cases": [
    {
      "id": "focus-switching",
      "category": "focus",
      "language": "en",
      "query": "How can I focus on one task without constantly switching?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "attention-focus"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "90-30-focus-rest-schedule",
          "ergonomic-workspace-assessment",
          "ready-to-resume-plan",
          "10-minute-language-microsprints",
          "2-minute-desk-tidy-timer"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "attention-focus",
          "task-initiation",
          "planning-prioritization"
        ],
        "protocol_slugs": [
          "ready-to-resume-plan",
          "90-30-focus-rest-schedule",
          "refocus-present-stop-mind-wandering",
          "batch-non-urgent-notifications",
          "30-minute-deadline-sprints"
        ],
        "evidence_decision_ids": [
          "ready-to-resume-interruption-boundary-2018",
          "notification-blocking-workday-boundary-2023",
          "microbreak-wellbeing-performance-boundary-2022"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            },
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "30-minute-deadline-sprints",
              "canonical_id": "brali:protocol:30-minute-deadline-sprints",
              "action": "Pick a low-risk task where an imperfect first pass is useful. Set a short timebox, define the minimum useful output, and decide what you will leave for later. Stop or reassess when the timebox ends.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "ready-to-resume-interruption-boundary-2018",
              "decision": "propose-protocol",
              "supported_claim": "When unfinished work must be interrupted, a short resumption plan can reduce attention residue and protect performance in the studied interruption contexts.",
              "limitations": [
                "The evidence comes from a small set of controlled interruption studies and does not represent every form of complex, collaborative or high-stakes real-world work.",
                "A ready-to-resume plan can mitigate attention residue in the studied contexts; it does not make task switching cost-free or imply that avoidable interruptions should be accepted.",
                "The research supports making a concrete resumption plan, not Brali's exact three-prompt note format, note length, writing medium or timing rule.",
                "The studies do not establish a universal productivity percentage, a guaranteed performance benefit, or the same effect for every individual and task."
              ],
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "id": "notification-blocking-workday-boundary-2023",
              "decision": "support-existing",
              "supported_claim": "Disabling automatic notifications can reduce notification interruptions; in this field experiment, fewer interruptions mediated better perceived performance and lower irritation.",
              "limitations": [
                "The study was a one-day field experiment, so it does not establish durable effects of notification blocking over longer work periods or adaptation over time.",
                "Performance and irritation outcomes were self-reported, and the direct intervention effect on performance was not significant; the reported performance pathway was indirect through fewer interruptions.",
                "Work roles differ in response-time obligations, telepressure, fear of missing out and safety or escalation requirements, so an always-off notification rule is not supported.",
                "The study tests communication-application notification interruptions, not every form of digital distraction or every focus-window design."
              ],
              "source_url": "https://doi.org/10.1002/1348-9585.12408"
            },
            {
              "id": "microbreak-wellbeing-performance-boundary-2022",
              "decision": "support-existing",
              "supported_claim": "Across the included studies, micro-breaks of up to 10 minutes were associated on average with small improvements in vigor and reduced fatigue. The pooled overall performance effect was not statistically significant, and performance effects varied by task demands and break duration. This supports retaining a short, adjustable break as a bounded recovery practice without presenting it as a reliable performance booster.",
              "limitations": [
                "The review combined heterogeneous break activities, durations, tasks and settings rather than testing a single Pomodoro-style intervention.",
                "Only four of twenty-two included study samples were rated low risk of bias across the assessed domains.",
                "The overall pooled performance effect was not statistically significant and performance heterogeneity was substantial.",
                "Performance effects differed by task type; the authors reported significant effects only for less cognitively demanding tasks in subgroup analyses.",
                "Longer break duration was associated with better performance in meta-regression, which argues against treating one short duration as universally optimal.",
                "Most outcomes were immediate post-break measures; the review does not establish long-term productivity or health effects.",
                "The source defines micro-breaks as no more than ten minutes and does not validate the work interval preceding the break."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9432722/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "pomodoro-exact",
      "category": "focus",
      "language": "en",
      "query": "Give me a 25 minute Pomodoro focus sprint",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "attention-focus"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "25-minute-pomodoro-focus-sprints"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "25-minute-pomodoro-focus-sprints",
          "4-minute-hiit-tabata-workout",
          "10-minute-morning-stretch-routine",
          "10-minute-stress-walk",
          "90-30-focus-rest-schedule"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "attention-focus"
        ],
        "protocol_slugs": [
          "25-minute-pomodoro-focus-sprints",
          "batch-non-urgent-notifications",
          "90-30-focus-rest-schedule",
          "ready-to-resume-plan",
          "refocus-present-stop-mind-wandering"
        ],
        "evidence_decision_ids": [
          "microbreak-wellbeing-performance-boundary-2022",
          "notification-blocking-workday-boundary-2023"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 2,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "25-minute-pomodoro-focus-sprints",
              "canonical_id": "brali:protocol:25-minute-pomodoro-focus-sprints",
              "action": "Pick one concrete outcome, work on it for one 25-minute focus block, take a short break, and then decide whether another block is useful. Treat 25 minutes as a starting setting, not a scientifically optimal duration.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "microbreak-wellbeing-performance-boundary-2022",
              "decision": "support-existing",
              "supported_claim": "Across the included studies, micro-breaks of up to 10 minutes were associated on average with small improvements in vigor and reduced fatigue. The pooled overall performance effect was not statistically significant, and performance effects varied by task demands and break duration. This supports retaining a short, adjustable break as a bounded recovery practice without presenting it as a reliable performance booster.",
              "limitations": [
                "The review combined heterogeneous break activities, durations, tasks and settings rather than testing a single Pomodoro-style intervention.",
                "Only four of twenty-two included study samples were rated low risk of bias across the assessed domains.",
                "The overall pooled performance effect was not statistically significant and performance heterogeneity was substantial.",
                "Performance effects differed by task type; the authors reported significant effects only for less cognitively demanding tasks in subgroup analyses.",
                "Longer break duration was associated with better performance in meta-regression, which argues against treating one short duration as universally optimal.",
                "Most outcomes were immediate post-break measures; the review does not establish long-term productivity or health effects.",
                "The source defines micro-breaks as no more than ten minutes and does not validate the work interval preceding the break."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9432722/"
            },
            {
              "id": "notification-blocking-workday-boundary-2023",
              "decision": "support-existing",
              "supported_claim": "Disabling automatic notifications can reduce notification interruptions; in this field experiment, fewer interruptions mediated better perceived performance and lower irritation.",
              "limitations": [
                "The study was a one-day field experiment, so it does not establish durable effects of notification blocking over longer work periods or adaptation over time.",
                "Performance and irritation outcomes were self-reported, and the direct intervention effect on performance was not significant; the reported performance pathway was indirect through fewer interruptions.",
                "Work roles differ in response-time obligations, telepressure, fear of missing out and safety or escalation requirements, so an always-off notification rule is not supported.",
                "The study tests communication-application notification interruptions, not every form of digital distraction or every focus-window design."
              ],
              "source_url": "https://doi.org/10.1002/1348-9585.12408"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "start-procrastination",
      "category": "focus",
      "language": "en",
      "query": "I keep procrastinating and cannot get started on an important task",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "task-initiation"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "ai-code-scaffold-test-loop",
          "temptation-bundling",
          "10-minute-language-microsprints",
          "5-minute-meditation-habit-tracker",
          "90-30-focus-rest-schedule"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "task-initiation",
          "attention-focus",
          "planning-prioritization"
        ],
        "protocol_slugs": [
          "ready-to-resume-plan",
          "temptation-bundling",
          "30-minute-deadline-sprints",
          "2-minute-desk-tidy-timer",
          "if-then-rules-productivity"
        ],
        "evidence_decision_ids": [
          "ready-to-resume-interruption-boundary-2018",
          "temptation-bundling-exercise-boundary-2020",
          "task-choice-response-latency-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "slug": "30-minute-deadline-sprints",
              "canonical_id": "brali:protocol:30-minute-deadline-sprints",
              "action": "Pick a low-risk task where an imperfect first pass is useful. Set a short timebox, define the minimum useful output, and decide what you will leave for later. Stop or reassess when the timebox ends.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "2-minute-desk-tidy-timer",
              "canonical_id": "brali:protocol:2-minute-desk-tidy-timer",
              "action": "Set a short timer and tidy one small part of your workspace. Make only obvious moves, then stop before tidying turns into a larger task.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "if-then-rules-productivity",
              "canonical_id": "brali:protocol:if-then-rules-productivity",
              "action": "Choose one repeated situation, describe the cue in observable terms, choose a small action you can safely perform when the cue appears, add a fallback for a common exception, use the rule when the situation occurs, and revise or retire it when it stops fitting the work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1080/10463283.2024.2334563"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "ready-to-resume-interruption-boundary-2018",
              "decision": "propose-protocol",
              "supported_claim": "When unfinished work must be interrupted, a short resumption plan can reduce attention residue and protect performance in the studied interruption contexts.",
              "limitations": [
                "The evidence comes from a small set of controlled interruption studies and does not represent every form of complex, collaborative or high-stakes real-world work.",
                "A ready-to-resume plan can mitigate attention residue in the studied contexts; it does not make task switching cost-free or imply that avoidable interruptions should be accepted.",
                "The research supports making a concrete resumption plan, not Brali's exact three-prompt note format, note length, writing medium or timing rule.",
                "The studies do not establish a universal productivity percentage, a guaranteed performance benefit, or the same effect for every individual and task."
              ],
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "id": "task-choice-response-latency-boundary-2026",
              "decision": "watch",
              "supported_claim": "In this specific automated manufacturing setting, giving workers task choice changed where time was spent rather than improving aggregate productivity: response time rose by about 80 seconds, accepted-task completion time fell by about 26 seconds, and machine productivity did not significantly change. This makes a hybrid choice-versus-delegation design worth further study: offer bounded choice where response latency is not critical, while preserving direct ownership for urgent tasks.",
              "limitations": [
                "Only two plants from one company were studied, so plant-level random allocation still leaves only two clusters and limits causal generalization.",
                "The post-treatment period was only three weeks.",
                "The assignment system could not perfectly encode worker skills, so improved matching and motivational autonomy effects cannot be fully disentangled.",
                "The tasks were short troubleshooting responses to machine interruptions, where response latency is unusually important.",
                "Productivity was measured at the machine level rather than directly for individual tasks or workers.",
                "The treatment group started from relatively high productivity, which may have limited detectable gains.",
                "The satisfaction result came from an auxiliary single-item survey and was acknowledged by the authors as anecdotal.",
                "Findings around task choice remain mixed in the broader literature, according to the authors' own discussion."
              ],
              "source_url": "https://journals.sagepub.com/doi/10.1177/10591478251400469"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "prioritize-week",
      "category": "focus",
      "language": "en",
      "query": "How do I prioritize my work for this week?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "planning-prioritization"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "cardio-health-daily-habits",
          "weekly-theme-learning-sprints",
          "2-minute-desk-tidy-timer",
          "25-minute-pomodoro-focus-sprints",
          "3-3-3-workday-planner"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "work-systems",
          "planning-prioritization",
          "revision-feedback"
        ],
        "protocol_slugs": [
          "deliberate-email-checking-windows",
          "3-3-3-workday-planner",
          "brag-doc-wins-hub",
          "ai-code-scaffold-test-loop",
          "ai-task-frontier-test"
        ],
        "evidence_decision_ids": [
          "email-checking-frequency-stress-boundary-2015"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "deliberate-email-checking-windows",
              "canonical_id": "brali:protocol:deliberate-email-checking-windows",
              "action": "Choose a small number of email windows that still meet the real response expectations of your work and personal life. Close or hide the inbox between those windows and disable nonessential email alerts. Keep a separate urgent channel for issues that genuinely cannot wait. Do not treat three checks per day as a universal rule; that was the experimental condition, not an optimized schedule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            },
            {
              "slug": "3-3-3-workday-planner",
              "canonical_id": "brali:protocol:3-3-3-workday-planner",
              "action": "Divide the workday into up to three broad blocks. Give each block one to three clear outcomes that fit the time you actually have. Replan when meetings, urgent work, or delays change the day.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "brag-doc-wins-hub",
              "canonical_id": "brali:protocol:brag-doc-wins-hub",
              "action": "After a meaningful piece of work, add a short entry with the situation, your contribution, the concrete result or current state, and one or two tags. During a review, retrieve entries that match the purpose and verify any figures or claims before reusing them.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-code-scaffold-test-loop",
              "canonical_id": "brali:protocol:ai-code-scaffold-test-loop",
              "action": "Define the behavior and constraints first. Ask AI for a small implementation. Read the diff, run the existing test suite, add tests for edge cases and failure paths, and inspect security or data-handling implications. Keep only code you can explain and maintain.",
              "evidence_state": "reviewed",
              "source_url": "https://arxiv.org/abs/2302.06590"
            },
            {
              "slug": "ai-task-frontier-test",
              "canonical_id": "brali:protocol:ai-task-frontier-test",
              "action": "Choose a small sample containing easy, ordinary, and awkward cases. Define what counts as acceptable before running the model. Compare AI output with the source or a trusted human result, record failure patterns, and decide which cases can be assisted, which need review, and which should stay human-first.",
              "evidence_state": "reviewed",
              "source_url": "https://aiinstitute.hbs.edu/navigating-the-jagged-technological-frontier/"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "email-checking-frequency-stress-boundary-2015",
              "decision": "propose-protocol",
              "supported_claim": "When role expectations allow it, checking email less frequently than one's normal pattern can reduce daily stress. A practical Brali implementation is to use deliberate email windows while keeping a separate route for genuinely urgent work. The study supports less-frequent checking, not three checks per day as an optimized rule.",
              "limitations": [
                "Checking frequency was self-reported rather than objectively logged.",
                "The study explored a broad set of outcomes, increasing the chance of isolated significant results.",
                "Direct causal evidence was strongest for stress; broader wellbeing links were indirect correlational analyses through stress.",
                "There was no passive measurement-only control condition.",
                "The experiment did not control the response expectations imposed by participants' workplaces and contacts."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "goals-control",
      "category": "focus",
      "language": "en",
      "query": "How can I turn a vague goal into actions I can control?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "goals"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "affirmation-habit-tracker",
          "coot-ai-growth-journal",
          "circles-of-control-planner",
          "woop-goal-planner",
          "5-whys-goal-clarity"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "goals",
          "emotion-regulation",
          "planning-prioritization"
        ],
        "protocol_slugs": [
          "visible-progress-monitoring",
          "achieve-multiple-goals-at-once",
          "woop-goal-planner",
          "if-then-rules-productivity",
          "5-whys-goal-clarity"
        ],
        "evidence_decision_ids": [
          "progress-monitoring-goal-attainment-meta-2016",
          "woop-mcii-academic-procrastination-rct-2026",
          "woop-mcii-goal-attainment-meta-analysis-2021"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "visible-progress-monitoring",
              "canonical_id": "brali:protocol:visible-progress-monitoring",
              "action": "Choose one goal where feedback can still change what you do. Pick one observable indicator that is close to the real outcome or behavior you care about. Record it at checkpoints that match the task, compare it with the target or previous checkpoint, and decide one adjustment. Prefer a small trace you will actually use over a dashboard full of decorative metrics.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/bul0000025"
            },
            {
              "slug": "achieve-multiple-goals-at-once",
              "canonical_id": "brali:protocol:achieve-multiple-goals-at-once",
              "action": "Name two priorities, brainstorm one concrete action that can serve both, and reject the combination if it makes either priority weaker or unclear.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "woop-goal-planner",
              "canonical_id": "brali:protocol:woop-goal-planner",
              "action": "Choose one meaningful and reasonably feasible wish. Describe the outcome you want, identify an internal obstacle that could derail you, then write: If I notice this obstacle, then I will take this specific action. Try the plan in real life and revise it if the cue or response is not useful.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.3389/fpsyg.2021.565202"
            },
            {
              "slug": "if-then-rules-productivity",
              "canonical_id": "brali:protocol:if-then-rules-productivity",
              "action": "Choose one repeated situation, describe the cue in observable terms, choose a small action you can safely perform when the cue appears, add a fallback for a common exception, use the rule when the situation occurs, and revise or retire it when it stops fitting the work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1080/10463283.2024.2334563"
            },
            {
              "slug": "5-whys-goal-clarity",
              "canonical_id": "brali:protocol:5-whys-goal-clarity",
              "action": "Write one goal or problem, ask why it matters, and continue asking why while each answer adds something useful. Stop when the answers become repetitive, speculative, or split into different causes. Then choose one next step or one assumption to check.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "progress-monitoring-goal-attainment-meta-2016",
              "decision": "support-existing",
              "supported_claim": "Increasing progress monitoring improves goal attainment on average; physically recording or reporting progress was associated with larger effects.",
              "limitations": [
                "The meta-analysis combines heterogeneous goals, populations and monitoring interventions, so the pooled effect does not identify one optimal metric, cadence or tracking interface.",
                "Larger effects associated with physically recording or reporting progress do not establish that public leaderboards, streaks, social accountability or any specific app feature is universally beneficial.",
                "The evidence supports increasing useful progress monitoring on average, not maximal monitoring; excessive or poorly chosen measurement can still create burden or metric gaming.",
                "The synthesis does not validate Brali's exact checkpoint wording or determine which proxy measure is best for a particular user's goal."
              ],
              "source_url": "https://doi.org/10.1037/bul0000025"
            },
            {
              "id": "woop-mcii-academic-procrastination-rct-2026",
              "decision": "support-existing",
              "supported_claim": "In this specific undergraduate academic-procrastination study, MCII reduced task aversiveness and improved willingness to initiate academic tasks relative to a positive-thinking control, with effects reported through one-week follow-up. Diary reports also provided a behavioral task-initiation signal. It provides narrow corroborating evidence for the obstacle-plus-plan mechanism in an academic setting.",
              "limitations": [
                "The study was single-site and restricted to first-year undergraduates at one private Chinese university.",
                "Twenty of 101 randomized participants were excluded for missing reports or intervention nonadherence, leaving 81 in the final sample (19.8% attrition).",
                "The main psychological outcomes were self-reported; task initiation was derived from participant diary reports rather than an objective academic-performance endpoint.",
                "The context was academic procrastination rather than general goal pursuit.",
                "The follow-up was one week, so long-term persistence is unknown.",
                "The study does not justify generalizing to broad life outcomes, workplaces, clinical treatment, or safety-critical decisions."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/41601124/"
            },
            {
              "id": "woop-mcii-goal-attainment-meta-analysis-2021",
              "decision": "support-existing",
              "supported_claim": "Across the included field-intervention studies, mental contrasting with implementation intentions was associated with a small-to-medium average improvement in goal attainment. The pooled random-effects estimate was Hedges' g=0.336, while publication-bias adjustment produced a smaller estimate. This supports WOOP/MCII as a bounded self-regulation option for feasible goals, not as a guarantee of individual goal success.",
              "limitations": [
                "The included studies were heterogeneous in population, goal domain, intervention delivery and outcome measurement.",
                "The meta-analysis reported medium heterogeneity and mixed publication-bias diagnostics; trim-and-fill yielded a smaller adjusted estimate.",
                "Two very large MOOC studies accounted for most participants and were excluded from moderator analyses because of their numerical dominance.",
                "The number of studies was limited for moderator inference, and the authors explicitly called for additional studies.",
                "An average standardized effect cannot predict whether the protocol will help one individual with one goal."
              ],
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2021.565202/full"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "habit-consistency",
      "category": "habits",
      "language": "en",
      "query": "How can I make a useful habit more consistent?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "habits-consistency"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "temptation-bundling",
          "25-minute-pomodoro-focus-sprints",
          "3-minute-sensory-mindfulness-act",
          "30-minute-deadline-sprints",
          "4-minute-hiit-tabata-workout"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "habits-consistency",
          "sleep-circadian",
          "attention-focus"
        ],
        "protocol_slugs": [
          "temptation-bundling",
          "visible-progress-monitoring",
          "if-then-rules-productivity",
          "coot-ai-growth-journal",
          "ideal-sleep-hours-finder"
        ],
        "evidence_decision_ids": [
          "temptation-bundling-exercise-boundary-2020",
          "progress-monitoring-goal-attainment-meta-2016"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 2,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "slug": "visible-progress-monitoring",
              "canonical_id": "brali:protocol:visible-progress-monitoring",
              "action": "Choose one goal where feedback can still change what you do. Pick one observable indicator that is close to the real outcome or behavior you care about. Record it at checkpoints that match the task, compare it with the target or previous checkpoint, and decide one adjustment. Prefer a small trace you will actually use over a dashboard full of decorative metrics.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/bul0000025"
            },
            {
              "slug": "if-then-rules-productivity",
              "canonical_id": "brali:protocol:if-then-rules-productivity",
              "action": "Choose one repeated situation, describe the cue in observable terms, choose a small action you can safely perform when the cue appears, add a fallback for a common exception, use the rule when the situation occurs, and revise or retire it when it stops fitting the work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1080/10463283.2024.2334563"
            },
            {
              "slug": "coot-ai-growth-journal",
              "canonical_id": "brali:protocol:coot-ai-growth-journal",
              "action": "State one concrete goal or decision, provide the relevant constraints and what you already know, ask the AI for a small set of distinct options with assumptions and failure modes, challenge anything vague, verify factual claims that matter, choose one reversible next action yourself, and record what would cause you to change course.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ideal-sleep-hours-finder",
              "canonical_id": "brali:protocol:ideal-sleep-hours-finder",
              "action": "Choose a bedtime-to-wake window that fits your obligations and gives you a reasonable opportunity for enough sleep. For a week or two, record only a few things: roughly when you tried to sleep, when you got up, whether the night was unusually disrupted, and how alert or sleepy you felt during the day. Look for repeated patterns, not single-night scores. Change one practical constraint at a time, such as moving bedtime earlier when your current schedule routinely leaves too little time for sleep.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34507028/"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "id": "progress-monitoring-goal-attainment-meta-2016",
              "decision": "support-existing",
              "supported_claim": "Increasing progress monitoring improves goal attainment on average; physically recording or reporting progress was associated with larger effects.",
              "limitations": [
                "The meta-analysis combines heterogeneous goals, populations and monitoring interventions, so the pooled effect does not identify one optimal metric, cadence or tracking interface.",
                "Larger effects associated with physically recording or reporting progress do not establish that public leaderboards, streaks, social accountability or any specific app feature is universally beneficial.",
                "The evidence supports increasing useful progress monitoring on average, not maximal monitoring; excessive or poorly chosen measurement can still create burden or metric gaming.",
                "The synthesis does not validate Brali's exact checkpoint wording or determine which proxy measure is best for a particular user's goal."
              ],
              "source_url": "https://doi.org/10.1037/bul0000025"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "decision-uncertainty",
      "category": "decisions",
      "language": "en",
      "query": "Show me a circles of control planner for a decision with an uncertain outcome",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "decision-making"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "circles-of-control-planner"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "circles-of-control-planner",
          "ab-test-learning-loop",
          "ask-for-feedback-tracker",
          "avoid-gamblers-fallacy-trust-the-odds",
          "consider-the-opposite"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "decision-making",
          "emotion-regulation",
          "negotiation"
        ],
        "protocol_slugs": [
          "circles-of-control-planner",
          "consider-the-opposite",
          "ab-test-learning-loop",
          "avoid-gamblers-fallacy-trust-the-odds",
          "ai-decision-alternatives"
        ],
        "evidence_decision_ids": [
          "debiasing-education-transfer-boundary-2025",
          "consider-opposite-social-judgment-boundary-1984",
          "custom-social-media-brake-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "circles-of-control-planner",
              "canonical_id": "brali:protocol:circles-of-control-planner",
              "action": "Write down one concern, separate the outcome from the actions you can take, convert anything you can influence into one concrete behavior, and choose the next useful action without pretending you control the final result.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "consider-the-opposite",
              "canonical_id": "brali:protocol:consider-the-opposite",
              "action": "Write your current conclusion in one sentence. Then ask: what evidence, mechanism, or alternative explanation could make the opposite conclusion reasonable? Generate a concrete alternative rather than telling yourself to be objective. Check whether your decision would change if that alternative were true. Use this on decisions where biased assimilation or one-sided hypothesis testing is a real risk; do not turn it into endless doubt about routine choices.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "slug": "ab-test-learning-loop",
              "canonical_id": "brali:protocol:ab-test-learning-loop",
              "action": "Choose one reversible decision, define what you want to learn, compare two approaches as fairly as you can, and record the outcome before deciding what to try next. Use proper experimental design when the result needs statistical confidence.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "avoid-gamblers-fallacy-trust-the-odds",
              "canonical_id": "brali:protocol:avoid-gamblers-fallacy-trust-the-odds",
              "action": "Describe the next event and the recent streak. Ask what carries information or state from earlier outcomes into the next one. If there is no connection, ignore the streak and use the same probability or decision rule. If a connection may exist, identify the relevant state and gather that information. Use a pre-defined stop rule for repeated high-cost decisions.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-decision-alternatives",
              "canonical_id": "brali:protocol:ai-decision-alternatives",
              "action": "Write your objective, constraints, and current leading option. Ask AI for materially different alternatives, missing criteria, and the strongest case against your favorite. Verify any external facts, then compare options using your own criteria and downside limits. Do not ask the model to hide the trade-off inside a single confident recommendation.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41598-024-60220-5"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "debiasing-education-transfer-boundary-2025",
              "decision": "support-existing",
              "supported_claim": "Debiasing education can produce a small average improvement on targeted bias tasks. This supports modest expectations for a consider-the-opposite check while making the boundary explicit: depth of learning and transfer to meaningful real-world decisions remain uncertain.",
              "limitations": [
                "All included studies were rated unclear or high risk of bias.",
                "There was some evidence of publication bias.",
                "Interventions, bias targets and outcome measures were highly heterogeneous.",
                "Transfer beyond explicitly trained tasks was limited or uncertain in many studies.",
                "The pooled effect represents diverse educational interventions rather than the consider-the-opposite strategy alone."
              ],
              "source_url": "https://www.nature.com/articles/s41562-025-02253-y"
            },
            {
              "id": "consider-opposite-social-judgment-boundary-1984",
              "decision": "propose-protocol",
              "supported_claim": "When a judgment is vulnerable to one-sided evidence processing, deliberately generating an opposed possibility can reduce bias on some tasks more effectively than simply telling oneself to be fair or unbiased. Brali can use this as a concrete pre-decision check while preserving the possibility that the original conclusion remains correct.",
              "limitations": [
                "Classic laboratory/social-judgment evidence from undergraduate samples.",
                "Only two focal task domains were tested in the original article.",
                "Long-term persistence and broad transfer were not established.",
                "The authors note that considering the opposite can in some circumstances overweight disconfirming evidence and create a different form of partiality.",
                "Demand characteristics and task-specific effects remain possible."
              ],
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "id": "custom-social-media-brake-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For an adult who already wants to spend less time in a specific social-media app, a short trial of a user-configured digital brake is reasonable: choose the target app, choose a personally meaningful session threshold, make continued use require an explicit quit-or-continue decision, and identify an alternative activity while retaining the ability to override the prompt. In this small randomized trial, the bundled intervention reduced time on the target app over the short study period. The source supports the package as a behavior-change experiment; it does not identify the full-screen checkpoint, customization, self-monitoring or any other component as the unique cause.",
              "limitations": [
                "The randomized sample was small at 70 participants, and the study did not reach its original recruitment target.",
                "Twenty-six percent did not complete week 3; the final model for the primary problematic-social-media-use outcome included 46 participants.",
                "Participants were iPhone users, mostly students or young professionals, and self-selected into a study about social-media self-regulation.",
                "The control group received no active comparator and participants could not be blinded.",
                "The intervention bundled several behavior-change components, so the study cannot isolate the causal contribution of the quit-or-continue checkpoint, customization, alternative activities, goals, feedback or self-monitoring.",
                "Some psychological outcomes were self-reported; problematic social-media use and self-efficacy did not show robust improvement.",
                "The intervention period was short and there was no long-term follow-up establishing durable habit change.",
                "One author was employed by Wellspent GmbH during the intervention period and another was a company cofounder."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13062480/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "prevent-errors",
      "category": "decisions",
      "language": "en",
      "query": "I keep making preventable mistakes. What kind of checklist or review can help?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "error-prevention"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "talk-about-fears-coach",
          "walking-meeting-assistant",
          "10-minute-language-microsprints",
          "affirmation-habit-tracker",
          "ai-task-frontier-test"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "decision-making",
          "error-prevention",
          "task-initiation"
        ],
        "protocol_slugs": [
          "ai-decision-alternatives",
          "ai-premortem-risk-planner",
          "ab-test-learning-loop",
          "circles-of-control-planner",
          "consider-the-opposite"
        ],
        "evidence_decision_ids": [
          "debiasing-education-transfer-boundary-2025",
          "structured-peer-feedback-provision-2025",
          "empathic-paraphrasing-immediate-emotion-boundary-2012"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ai-decision-alternatives",
              "canonical_id": "brali:protocol:ai-decision-alternatives",
              "action": "Write your objective, constraints, and current leading option. Ask AI for materially different alternatives, missing criteria, and the strongest case against your favorite. Verify any external facts, then compare options using your own criteria and downside limits. Do not ask the model to hide the trade-off inside a single confident recommendation.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41598-024-60220-5"
            },
            {
              "slug": "ai-premortem-risk-planner",
              "canonical_id": "brali:protocol:ai-premortem-risk-planner",
              "action": "Describe the plan without sensitive information, ask an AI for plausible ways it could fail, group the suggestions, verify the important ones yourself, and choose mitigations only for risks that survive the check.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ab-test-learning-loop",
              "canonical_id": "brali:protocol:ab-test-learning-loop",
              "action": "Choose one reversible decision, define what you want to learn, compare two approaches as fairly as you can, and record the outcome before deciding what to try next. Use proper experimental design when the result needs statistical confidence.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "circles-of-control-planner",
              "canonical_id": "brali:protocol:circles-of-control-planner",
              "action": "Write down one concern, separate the outcome from the actions you can take, convert anything you can influence into one concrete behavior, and choose the next useful action without pretending you control the final result.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "consider-the-opposite",
              "canonical_id": "brali:protocol:consider-the-opposite",
              "action": "Write your current conclusion in one sentence. Then ask: what evidence, mechanism, or alternative explanation could make the opposite conclusion reasonable? Generate a concrete alternative rather than telling yourself to be objective. Check whether your decision would change if that alternative were true. Use this on decisions where biased assimilation or one-sided hypothesis testing is a real risk; do not turn it into endless doubt about routine choices.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "debiasing-education-transfer-boundary-2025",
              "decision": "support-existing",
              "supported_claim": "Debiasing education can produce a small average improvement on targeted bias tasks. This supports modest expectations for a consider-the-opposite check while making the boundary explicit: depth of learning and transfer to meaningful real-world decisions remain uncertain.",
              "limitations": [
                "All included studies were rated unclear or high risk of bias.",
                "There was some evidence of publication bias.",
                "Interventions, bias targets and outcome measures were highly heterogeneous.",
                "Transfer beyond explicitly trained tasks was limited or uncertain in many studies.",
                "The pooled effect represents diverse educational interventions rather than the consider-the-opposite strategy alone."
              ],
              "source_url": "https://www.nature.com/articles/s41562-025-02253-y"
            },
            {
              "id": "structured-peer-feedback-provision-2025",
              "decision": "support-existing",
              "supported_claim": "In the reviewed educational literature, adding instructional support to peer feedback improved the peer-feedback process overall, and support aimed at the person providing feedback was associated with better feedback-provision quality. This supports Brali's existing educational writing protocol in making the review target explicit and giving the peer reviewer a small amount of structure instead of requesting vague general feedback.",
              "limitations": [
                "Only 32 journal studies met the inclusion criteria, limiting fine-grained moderator analysis.",
                "Only three studies with 14 effect sizes examined feedback-reception support, making conclusions about that phase underpowered and unstable.",
                "The authors had to collapse specific supports such as rubrics, sentence starters and guiding questions into broader categories because each was represented by too few studies.",
                "Study heterogeneity was high, and publication-bias tests were mixed: Egger's test was significant whereas Begg's test was not and the funnel plot was relatively symmetrical.",
                "The meta-analysis concerns educational peer-feedback processes and should not be treated as direct evidence for all professional writing or workplace review contexts.",
                "Coarse outcome categories do not reveal which exact support mechanism is best for a specific writing task."
              ],
              "source_url": "https://link.springer.com/article/10.1007/s10648-025-10017-3"
            },
            {
              "id": "empathic-paraphrasing-immediate-emotion-boundary-2012",
              "decision": "support-existing",
              "supported_claim": "In this small nonclinical social-conflict experiment, a trained interviewer used a behavior closely matching Brali's core correction loop: summarize the speaker's facts, feelings and priorities, then ask whether the understanding is accurate. Participants reported less negative immediate emotion after paraphrasing than after silent note-taking. This supports retaining paraphrase plus explicit correction as a bounded practice behavior, with the studied outcome and setting stated precisely.",
              "limitations": [
                "Only twenty participants contributed self-report data, making results vulnerable to individual variation and unsuitable for broad population estimates.",
                "One female interviewer with approximately 190 hours of conflict-resolution training delivered every paraphrase, so listener skill and delivery cannot be separated from the technique.",
                "The control condition was silent note-taking rather than another spoken response; differences may partly reflect receiving a verbal response rather than paraphrasing specifically.",
                "Participants may have interpreted note-taking as judgment despite the study explanation, potentially biasing the comparison.",
                "Only immediate reactions were measured; the authors explicitly described longer-term emotion resolution as speculative.",
                "Conflicts involving physical or psychological violence were excluded, so the result must not be applied as ordinary advice in unsafe or abusive situations.",
                "The study assessed emotional valence and arousal, not whether the listener understood more accurately or whether the conflict outcome improved."
              ],
              "source_url": "https://doi.org/10.3389/fpsyg.2012.00482"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "stress-overload",
      "category": "stress",
      "language": "en",
      "query": "Show me a short stress walk I can use as a low-risk reset after work",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "stress-regulation"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "10-minute-stress-walk"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "30-minute-deadline-sprints",
          "10-minute-stress-walk",
          "2-minute-desk-tidy-timer",
          "4-minute-tabata-hiit-timer",
          "ai-rubric-critique"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "stress-regulation",
          "work-systems",
          "planning-prioritization"
        ],
        "protocol_slugs": [
          "10-minute-stress-walk",
          "abdominal-breathing-stress-relief",
          "dbt-distress-tolerance-coach",
          "3-minute-sensory-mindfulness-act",
          "ancestral-torch-resilience-tracker"
        ],
        "evidence_decision_ids": [
          "green-route-stress-walk-boundary-2026",
          "activity-breaks-postprandial-metabolism-boundary-2026",
          "microbreak-wellbeing-performance-boundary-2022"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "10-minute-stress-walk",
              "canonical_id": "brali:protocol:10-minute-stress-walk",
              "action": "When stress is building, consider a short, comfortable walk if walking is safe and practical for you. Treat ten minutes as a convenient boundary, not a medically proven dose. If a greener route is just as easy and safe, prefer it as a reasonable experiment rather than making a special trip. Then notice whether you feel any different before returning to the next task.",
              "evidence_state": "reviewed",
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "slug": "abdominal-breathing-stress-relief",
              "canonical_id": "brali:protocol:abdominal-breathing-stress-relief",
              "action": "Sit or lie in a comfortable position, let the breath slow without forcing a count, notice the abdomen rise and fall, and stop after a few easy cycles or whenever you prefer.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nccih.nih.gov/health/relaxation-techniques-what-you-need-to-know"
            },
            {
              "slug": "dbt-distress-tolerance-coach",
              "canonical_id": "brali:protocol:dbt-distress-tolerance-coach",
              "action": "Pause. Describe the situation in one factual sentence and name the action urge without judging it. Choose a harmless brief activity that creates time rather than escalating the situation. Then reassess what action is safest and most useful.",
              "evidence_state": "reviewed",
              "source_url": "https://www.dorsethealthcare.nhs.uk/our-services-and-sites/mental-health-and-learning-disabilities/intensive-psychological-therapies/dialectical-behaviour-therapy-dbt"
            },
            {
              "slug": "3-minute-sensory-mindfulness-act",
              "canonical_id": "brali:protocol:3-minute-sensory-mindfulness-act",
              "action": "Pause somewhere safe and deliberately notice what is happening around you right now. Choose a few sights, sounds, textures, smells or tastes and describe them simply. When your attention drifts into planning, worry or commentary, notice that and gently return to one sensory detail.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/every-mind-matters/mental-wellbeing-tips/what-is-mindfulness/"
            },
            {
              "slug": "ancestral-torch-resilience-tracker",
              "canonical_id": "brali:protocol:ancestral-torch-resilience-tracker",
              "action": "Write one sentence in the form: This is difficult, and the next thing I can control is ___. Keep it specific to the current situation.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "green-route-stress-walk-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "For an existing Brali stress-walk protocol, a safe greener route can be offered as a low-friction optional modifier when it is as convenient as an ordinary route. In the reviewed randomized evidence, green exercise improved average wellbeing and affect relative to pooled non-exercise, indoor-exercise and built-up outdoor comparators. Stress-specific pooled effects favored green exercise in all three comparator categories. This supports 'prefer green when equally practical', not a stronger prescription.",
              "limitations": [
                "Several pooled outcomes showed substantial or high heterogeneity, although sensitivity analyses suggested some results were driven by a small number of studies.",
                "Most individual studies were small, and the authors noted incomplete reporting of randomization procedures and limited blinding in some trials.",
                "Most interventions were walking, so generalization to other forms or intensities of exercise is limited.",
                "Most interventions were short-term, with follow-up generally under three months, so long-term durability is uncertain.",
                "Settings, exercise formats and psychological measures varied across studies.",
                "Stress-specific comparisons contained relatively few studies: three versus non-exercise, six versus indoor exercise and two versus built-up exercise.",
                "The meta-analysis does not isolate a unique biological or psychological mechanism for any observed advantage of greener settings."
              ],
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "id": "activity-breaks-postprandial-metabolism-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For acute post-meal metabolism, replacing small portions of a prolonged sitting period with brief activity is better supported than remaining continuously seated. Across randomized crossover evidence, regular activity breaks reduced postprandial glucose and insulin responses, with walking producing the strongest overall mode estimates. Brali can therefore propose a bounded protocol to interrupt long sitting bouts with brief movement, especially walking when feasible, while treating the exact timing and dose as adaptable rather than universally fixed.",
              "limitations": [
                "All included studies were acute laboratory randomized crossover trials lasting less than 24 hours, so the evidence does not establish long-term health effects or sustainability.",
                "Prolonged-sitting laboratory protocols, especially the longer ones, may not resemble habitual real-world sitting.",
                "Many pooled and subgroup estimates had substantial heterogeneity that remained after subgroup analysis.",
                "Most studies were rated fair or good rather than excellent; participant blinding was impossible and reporting of attrition and assessor blinding was inconsistent.",
                "The review was not prospectively registered in PROSPERO or another registry and no review protocol was prepared.",
                "The search was limited to peer-reviewed English-language studies, so relevant evidence may have been missed.",
                "Frequency and mode subgroup comparisons do not by themselves prove that the largest subgroup estimate is the optimal prescription for an individual."
              ],
              "source_url": "https://onlinelibrary.wiley.com/doi/10.1111/obr.70152"
            },
            {
              "id": "microbreak-wellbeing-performance-boundary-2022",
              "decision": "support-existing",
              "supported_claim": "Across the included studies, micro-breaks of up to 10 minutes were associated on average with small improvements in vigor and reduced fatigue. The pooled overall performance effect was not statistically significant, and performance effects varied by task demands and break duration. This supports retaining a short, adjustable break as a bounded recovery practice without presenting it as a reliable performance booster.",
              "limitations": [
                "The review combined heterogeneous break activities, durations, tasks and settings rather than testing a single Pomodoro-style intervention.",
                "Only four of twenty-two included study samples were rated low risk of bias across the assessed domains.",
                "The overall pooled performance effect was not statistically significant and performance heterogeneity was substantial.",
                "Performance effects differed by task type; the authors reported significant effects only for less cognitively demanding tasks in subgroup analyses.",
                "Longer break duration was associated with better performance in meta-regression, which argues against treating one short duration as universally optimal.",
                "Most outcomes were immediate post-break measures; the review does not establish long-term productivity or health effects.",
                "The source defines micro-breaks as no more than ten minutes and does not validate the work interval preceding the break."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC9432722/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "emotion-reactivity",
      "category": "stress",
      "language": "en",
      "query": "How can I notice a strong emotion before reacting automatically?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "emotion-regulation"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "dbt-distress-tolerance-coach",
          "10-minute-stress-walk",
          "3-minute-sensory-mindfulness-act",
          "5-minute-meditation-habit-tracker",
          "ab-test-learning-loop"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "emotion-regulation",
          "self-awareness"
        ],
        "protocol_slugs": [
          "dbt-distress-tolerance-coach",
          "cbt-action-mood-tracker",
          "3-minute-sensory-mindfulness-act",
          "daily-gratitude-journal-three-things",
          "savor-positive-moment"
        ],
        "evidence_decision_ids": [
          "savoring-emotional-outcomes-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "dbt-distress-tolerance-coach",
              "canonical_id": "brali:protocol:dbt-distress-tolerance-coach",
              "action": "Pause. Describe the situation in one factual sentence and name the action urge without judging it. Choose a harmless brief activity that creates time rather than escalating the situation. Then reassess what action is safest and most useful.",
              "evidence_state": "reviewed",
              "source_url": "https://www.dorsethealthcare.nhs.uk/our-services-and-sites/mental-health-and-learning-disabilities/intensive-psychological-therapies/dialectical-behaviour-therapy-dbt"
            },
            {
              "slug": "cbt-action-mood-tracker",
              "canonical_id": "brali:protocol:cbt-action-mood-tracker",
              "action": "Write the situation, the activity you chose, how you felt before, and how you felt afterward. Add one alternative explanation so the note does not turn correlation into certainty.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/tests-and-treatments/cognitive-behavioural-therapy-cbt/"
            },
            {
              "slug": "3-minute-sensory-mindfulness-act",
              "canonical_id": "brali:protocol:3-minute-sensory-mindfulness-act",
              "action": "Pause somewhere safe and deliberately notice what is happening around you right now. Choose a few sights, sounds, textures, smells or tastes and describe them simply. When your attention drifts into planning, worry or commentary, notice that and gently return to one sensory detail.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/every-mind-matters/mental-wellbeing-tips/what-is-mindfulness/"
            },
            {
              "slug": "daily-gratitude-journal-three-things",
              "canonical_id": "brali:protocol:daily-gratitude-journal-three-things",
              "action": "Choose one to three concrete moments, people, resources or actions from the day. Write what happened and one sentence about why it mattered to you. Skip the practice when it feels forced or unhelpful.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/40627390/"
            },
            {
              "slug": "savor-positive-moment",
              "canonical_id": "brali:protocol:savor-positive-moment",
              "action": "Choose a positive experience that is already real: a good meal, a joke, a quiet walk, finishing useful work, music, a kind message, or another ordinary moment. For a short period, stop adding new input. Notice one or two concrete details and what makes the moment worth noticing. Then continue your day. Do not use the exercise to deny an unpleasant feeling or to manufacture gratitude for something that is not good.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1111/aphw.70134"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "savoring-emotional-outcomes-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "Savoring interventions improve emotional outcomes on average across randomized trials, but the evidence does not identify one optimal micro-practice. Brali can reasonably offer a minimal nonclinical implementation—deliberately attending to an already-positive experience—as a personal experiment while clearly labeling the concrete format as an implementation choice rather than the meta-analysis's tested universal recipe.",
              "limitations": [
                "Total heterogeneity was high, with I2 86.61% across effects.",
                "Six of 20 studies were rated high risk of bias and ten had some concerns.",
                "Most outcomes were self-reported and therefore vulnerable to recall and social-desirability bias.",
                "Intervention content, delivery format, populations and duration varied substantially.",
                "The median intervention duration was about 14 days, limiting long-term conclusions.",
                "All included samples were adults, limiting generalization to children and adolescents.",
                "Active-control point estimates were smaller than passive-control estimates, although control type did not reach statistical significance as a moderator."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12968602/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "fear-presentation",
      "category": "stress",
      "language": "en",
      "query": "How can I build confidence for a presentation I am afraid of?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "fear-confidence"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "10-minute-morning-stretch-routine",
          "10-minute-stress-walk",
          "ab-test-learning-loop",
          "automate-routine-data-reports",
          "brainwriting-group-idea-generation"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "fear-confidence",
          "social-confidence",
          "metacognition"
        ],
        "protocol_slugs": [
          "talk-about-fears-coach",
          "safety-statements-exposure-anxiety",
          "genuine-micro-conversation",
          "ai-learning-attempt-feedback-retest",
          "prequestions-before-learning"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "talk-about-fears-coach",
              "canonical_id": "brali:protocol:talk-about-fears-coach",
              "action": "Choose someone you trust and tell them, in your own words, what you are worried or afraid about. You can also say what kind of support you want, such as listening, practical help, or simply company. Keep the conversation within boundaries that feel appropriate for both of you.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/every-mind-matters/mental-wellbeing-tips/how-to-talk-about-your-mental-health/"
            },
            {
              "slug": "safety-statements-exposure-anxiety",
              "canonical_id": "brali:protocol:safety-statements-exposure-anxiety",
              "action": "Choose one situation you avoid because it feels frightening but that is not actually dangerous. Break it into smaller steps, start with a manageable step, approach it gradually, and notice what happens rather than trying to force fear away with a required phrase or ritual.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/every-mind-matters/mental-wellbeing-tips/self-help-cbt-techniques/facing-your-fears/"
            },
            {
              "slug": "genuine-micro-conversation",
              "canonical_id": "brali:protocol:genuine-micro-conversation",
              "action": "Pick an ordinary low-pressure interaction that already needs to happen: buying coffee, greeting a neighbor, talking to a receptionist, or another brief exchange. If the context is appropriate, make eye contact, greet the person, and add one natural sentence or question. Keep it short and responsive to the other person's cues. If they are busy, uncomfortable, or not interested, let the interaction stay efficient.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1177/1948550613502990"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "worry-loop",
      "category": "stress",
      "language": "en",
      "query": "I keep replaying the same worry in my head. How can I redirect attention?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "worry-rumination"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "3-minute-sensory-mindfulness-act",
          "refocus-present-stop-mind-wandering",
          "temptation-bundling",
          "5-minute-meditation-habit-tracker",
          "oblique-strategies-prompt-cards"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "worry-rumination",
          "attention-focus",
          "digital-attention"
        ],
        "protocol_slugs": [
          "batch-non-urgent-notifications",
          "refocus-present-stop-mind-wandering",
          "ready-to-resume-plan",
          "90-30-focus-rest-schedule",
          "deliberate-email-checking-windows"
        ],
        "evidence_decision_ids": [
          "notification-batching-boundary-2019",
          "ready-to-resume-interruption-boundary-2018",
          "email-checking-frequency-stress-boundary-2015"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "deliberate-email-checking-windows",
              "canonical_id": "brali:protocol:deliberate-email-checking-windows",
              "action": "Choose a small number of email windows that still meet the real response expectations of your work and personal life. Close or hide the inbox between those windows and disable nonessential email alerts. Keep a separate urgent channel for issues that genuinely cannot wait. Do not treat three checks per day as a universal rule; that was the experimental condition, not an optimized schedule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "notification-batching-boundary-2019",
              "decision": "propose-protocol",
              "supported_claim": "Predictable batching of non-urgent notifications is a defensible attention-environment experiment. In this field trial, three daily batches reduced perceived interruption and stress relative to usual delivery, while hourly batching changed little and complete notification removal increased anxiety/FoMO. Brali should preserve urgent exceptions and treat schedule frequency as a user-fit parameter rather than a fixed dose.",
              "limitations": [
                "Single short field experiment rather than a replicated long-term evidence base.",
                "Participants were recruited through an online labor market and may not represent all smartphone users or work contexts.",
                "The intervention used a custom notification-management implementation.",
                "Many psychological outcomes were self-reported and the study intentionally used broad exploratory measurement.",
                "The study compared a small set of schedules and cannot identify an optimal individualized batching frequency."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "id": "ready-to-resume-interruption-boundary-2018",
              "decision": "propose-protocol",
              "supported_claim": "When unfinished work must be interrupted, a short resumption plan can reduce attention residue and protect performance in the studied interruption contexts.",
              "limitations": [
                "The evidence comes from a small set of controlled interruption studies and does not represent every form of complex, collaborative or high-stakes real-world work.",
                "A ready-to-resume plan can mitigate attention residue in the studied contexts; it does not make task switching cost-free or imply that avoidable interruptions should be accepted.",
                "The research supports making a concrete resumption plan, not Brali's exact three-prompt note format, note length, writing medium or timing rule.",
                "The studies do not establish a universal productivity percentage, a guaranteed performance benefit, or the same effect for every individual and task."
              ],
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "id": "email-checking-frequency-stress-boundary-2015",
              "decision": "propose-protocol",
              "supported_claim": "When role expectations allow it, checking email less frequently than one's normal pattern can reduce daily stress. A practical Brali implementation is to use deliberate email windows while keeping a separate route for genuinely urgent work. The study supports less-frequent checking, not three checks per day as an optimized rule.",
              "limitations": [
                "Checking frequency was self-reported rather than objectively logged.",
                "The study explored a broad set of outcomes, increasing the chance of isolated significant results.",
                "Direct causal evidence was strongest for stress; broader wellbeing links were indirect correlational analyses through stress.",
                "There was no passive measurement-only control condition.",
                "The experiment did not control the response expectations imposed by participants' workplaces and contacts."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "sleep-routine",
      "category": "sleep",
      "language": "en",
      "query": "How can I make my sleep schedule more regular?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "sleep-circadian"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "ideal-sleep-hours-finder",
          "20-20-20-eye-break-reminder",
          "90-30-focus-rest-schedule",
          "automatic-paycheck-to-savings",
          "batch-non-urgent-notifications"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "sleep-circadian",
          "digital-wellbeing"
        ],
        "protocol_slugs": [
          "ideal-sleep-hours-finder",
          "stop-caffeine-after-lunch",
          "20-20-20-eye-break-reminder"
        ],
        "evidence_decision_ids": [
          "sleep-opportunity-extension-2021",
          "caffeine-dose-timing-sleep-boundary-2025"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 2,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ideal-sleep-hours-finder",
              "canonical_id": "brali:protocol:ideal-sleep-hours-finder",
              "action": "Choose a bedtime-to-wake window that fits your obligations and gives you a reasonable opportunity for enough sleep. For a week or two, record only a few things: roughly when you tried to sleep, when you got up, whether the night was unusually disrupted, and how alert or sleepy you felt during the day. Look for repeated patterns, not single-night scores. Change one practical constraint at a time, such as moving bedtime earlier when your current schedule routinely leaves too little time for sleep.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34507028/"
            },
            {
              "slug": "stop-caffeine-after-lunch",
              "canonical_id": "brali:protocol:stop-caffeine-after-lunch",
              "action": "For one or two weeks, note the time and rough amount of your last caffeine, your bedtime, and whether sleep onset, overnight restfulness, and next-day sleepiness are broadly better or worse. Start by moving the last large dose earlier or reducing it; do not deliberately take caffeine late just to test your tolerance.",
              "evidence_state": "reviewed",
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC11985402/"
            },
            {
              "slug": "20-20-20-eye-break-reminder",
              "canonical_id": "brali:protocol:20-20-20-eye-break-reminder",
              "action": "During long periods of screen use, take a short visual break about every 20 minutes and look at something about 20 feet away for 20 seconds. Use the reminder as a simple screen-rest cue, not as a treatment or a substitute for eye care.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nei.nih.gov/eye-health-information/healthy-vision/how-eyes-work/keep-your-eyes-healthy"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "sleep-opportunity-extension-2021",
              "decision": "propose-protocol",
              "supported_claim": "Behavioral interventions designed to extend sleep increased sleep duration on average compared with control or baseline. Direct interventions that specified a sleep schedule tended to have larger effects, but results were highly heterogeneous.",
              "limitations": [
                "Statistical heterogeneity was very high across both two-arm and one-arm studies.",
                "Populations, intervention components, duration and measurement methods varied substantially.",
                "The primary outcome was sleep duration; the review does not establish a single personal optimum for daytime performance or health.",
                "The findings do not replace clinical assessment for persistent insomnia, excessive sleepiness, suspected sleep disorders, or other medical concerns."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34507028/"
            },
            {
              "id": "caffeine-dose-timing-sleep-boundary-2025",
              "decision": "support-existing",
              "supported_claim": "Caffeine timing should be interpreted together with dose rather than as one universal after-lunch rule. In this small randomized crossover trial, a single 400 mg dose disrupted multiple objective sleep measures when taken within 12 hours of bedtime, with larger disruption closer to bedtime; no statistically significant sleep effect was detected for 100 mg at the tested 12-, 8-, and 4-hour timings. For Brali, this supports moving large late-day caffeine doses earlier when sleep matters and using bedtime-relative timing rather than a fixed clock cutoff.",
              "limitations": [
                "The sample was small (23 participants) and included only healthy men aged 18–40, limiting generalizability to women, older adults, adolescents, people with sleep disorders, and other populations.",
                "Participants were moderate habitual caffeine users; results may differ in people with very low, very high, or irregular caffeine exposure.",
                "The intervention used acute standardized caffeine capsules rather than the variable doses, ingredients, absorption patterns, and repeated servings found in real-world caffeinated products.",
                "The study tested only two caffeine doses and three pre-bed timing points, so it cannot define a continuous or personalized cutoff.",
                "Sleep was measured with in-home partial polysomnography rather than full laboratory polysomnography.",
                "A nonsignificant result for 100 mg in this sample is not proof of no effect for every individual, especially given known variability in caffeine pharmacokinetics and sensitivity."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC11985402/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "movement-sitting",
      "category": "movement",
      "language": "en",
      "query": "I sit at a screen for hours. Show me a 20 20 20 eye break reminder for recovery.",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "recovery-energy"
        ],
        "acceptable_topic_ids": [
          "digital-attention",
          "movement"
        ],
        "protocol_slugs": [
          "20-20-20-eye-break-reminder"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "20-20-20-eye-break-reminder",
          "30-min-eye-break-tracker",
          "4-minute-tabata-hiit-timer",
          "walking-meeting-assistant",
          "10-minute-morning-stretch-routine"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "movement",
          "recovery-energy",
          "digital-wellbeing"
        ],
        "protocol_slugs": [
          "20-20-20-eye-break-reminder",
          "post-meal-walk",
          "10-minute-stress-walk",
          "cardio-health-daily-habits",
          "temptation-bundling"
        ],
        "evidence_decision_ids": [
          "post-meal-exercise-acute-glucose-boundary-2023",
          "green-route-stress-walk-boundary-2026",
          "temptation-bundling-exercise-boundary-2020"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "20-20-20-eye-break-reminder",
              "canonical_id": "brali:protocol:20-20-20-eye-break-reminder",
              "action": "During long periods of screen use, take a short visual break about every 20 minutes and look at something about 20 feet away for 20 seconds. Use the reminder as a simple screen-rest cue, not as a treatment or a substitute for eye care.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nei.nih.gov/eye-health-information/healthy-vision/how-eyes-work/keep-your-eyes-healthy"
            },
            {
              "slug": "post-meal-walk",
              "canonical_id": "brali:protocol:post-meal-walk",
              "action": "After a meal, choose an ordinary walk that feels comfortable and fits the situation. Starting relatively soon after eating matches the direction of the reviewed acute evidence better than deliberately waiting a long time, but Brali does not prescribe a universal minute mark, pace, or duration. If you use glucose-lowering medication, have a condition affected by exercise or meals, or have been given specific activity advice, follow your clinical plan instead of using this page as treatment guidance.",
              "evidence_state": "reviewed",
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC10036272/"
            },
            {
              "slug": "10-minute-stress-walk",
              "canonical_id": "brali:protocol:10-minute-stress-walk",
              "action": "When stress is building, consider a short, comfortable walk if walking is safe and practical for you. Treat ten minutes as a convenient boundary, not a medically proven dose. If a greener route is just as easy and safe, prefer it as a reasonable experiment rather than making a special trip. Then notice whether you feel any different before returning to the next task.",
              "evidence_state": "reviewed",
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "slug": "cardio-health-daily-habits",
              "canonical_id": "brali:protocol:cardio-health-daily-habits",
              "action": "Look at the coming week, mark the activity you already do, then add realistic walking, moderate or vigorous activity, and strength sessions where they fit. Build gradually rather than treating the guideline as a one-day target.",
              "evidence_state": "reviewed",
              "source_url": "https://www.heart.org/en/healthy-living/healthy-lifestyle/lifes-essential-8/how-to-be-more-active-fact-sheet"
            },
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "post-meal-exercise-acute-glucose-boundary-2023",
              "decision": "propose-protocol",
              "supported_claim": "Replacing some post-meal sitting with safe walking or other suitable movement relatively soon after eating is a defensible practical option when the goal is ordinary movement and the acute post-meal glucose mechanism is relevant. The meta-analysis found lower acute postprandial glucose after post-meal exercise than after no exercise and better average results than matched pre-meal exercise. Brali should not prescribe one duration, pace, start minute, or medical target.",
              "limitations": [
                "Only eight randomized crossover trials with 116 total participants met the strict inclusion criteria.",
                "Included trials were rated high risk of bias.",
                "Exercise protocols, meals, participant characteristics and glucose measurement methods varied.",
                "The type 2 diabetes subgroup was small and some subgroup comparisons were imprecise.",
                "The review concerned acute postprandial responses and cannot establish long-term clinical outcomes.",
                "Moderator evidence about delay after eating does not define an exact individual optimum."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC10036272/"
            },
            {
              "id": "green-route-stress-walk-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "For an existing Brali stress-walk protocol, a safe greener route can be offered as a low-friction optional modifier when it is as convenient as an ordinary route. In the reviewed randomized evidence, green exercise improved average wellbeing and affect relative to pooled non-exercise, indoor-exercise and built-up outdoor comparators. Stress-specific pooled effects favored green exercise in all three comparator categories. This supports 'prefer green when equally practical', not a stronger prescription.",
              "limitations": [
                "Several pooled outcomes showed substantial or high heterogeneity, although sensitivity analyses suggested some results were driven by a small number of studies.",
                "Most individual studies were small, and the authors noted incomplete reporting of randomization procedures and limited blinding in some trials.",
                "Most interventions were walking, so generalization to other forms or intensities of exercise is limited.",
                "Most interventions were short-term, with follow-up generally under three months, so long-term durability is uncertain.",
                "Settings, exercise formats and psychological measures varied across studies.",
                "Stress-specific comparisons contained relatively few studies: three versus non-exercise, six versus indoor exercise and two versus built-up exercise.",
                "The meta-analysis does not isolate a unique biological or psychological mechanism for any observed advantage of greener settings."
              ],
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "stretch-exact",
      "category": "movement",
      "language": "en",
      "query": "Show me a 10 minute morning stretch routine",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [],
        "acceptable_topic_ids": [
          "fitness-strength",
          "movement"
        ],
        "protocol_slugs": [
          "10-minute-morning-stretch-routine"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "10-minute-morning-stretch-routine",
          "10-minute-stress-walk",
          "4-minute-hiit-tabata-workout",
          "25-minute-pomodoro-focus-sprints",
          "4-minute-tabata-hiit-timer"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "fitness-strength"
        ],
        "protocol_slugs": [
          "10-minute-morning-stretch-routine",
          "4-minute-hiit-tabata-workout",
          "4-minute-tabata-hiit-timer",
          "10-minute-stress-walk"
        ],
        "evidence_decision_ids": [
          "exercise-snacks-real-world-t2d-boundary-2026",
          "exercise-snacks-cardiorespiratory-fitness-boundary-2026",
          "green-route-stress-walk-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "10-minute-morning-stretch-routine",
              "canonical_id": "brali:protocol:10-minute-morning-stretch-routine",
              "action": "Use a short, gentle flexibility routine as a morning movement cue if that timing suits you. Move slowly only as far as is comfortable, start with a few simple movements, and build up gradually rather than forcing a fixed 10-to-20-minute progression.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/live-well/exercise/flexibility-exercises/"
            },
            {
              "slug": "4-minute-hiit-tabata-workout",
              "canonical_id": "brali:protocol:4-minute-hiit-tabata-workout",
              "action": "If vigorous exercise is appropriate for your current fitness and health, choose a simple movement you can perform safely. You can use it as one short interval session, or as brief exercise snacks separated across the day when a full workout is unlikely. Make the effort challenging but controlled. Treat 20/10, 3 × 1 minute, 4 × 1 minute and similar timers as formats to adapt, not scientifically optimized doses.",
              "evidence_state": "reviewed",
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13486332/"
            },
            {
              "slug": "4-minute-tabata-hiit-timer",
              "canonical_id": "brali:protocol:4-minute-tabata-hiit-timer",
              "action": "If very vigorous exercise is suitable for you, use an interval timer to alternate a short period of harder work with recovery. A 20-second work and 10-second recovery pattern can be one option, but choose a movement and number of rounds you can control safely rather than treating 8 all-out rounds as mandatory.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/live-well/exercise/physical-activity-guidelines-for-adults-aged-19-to-64/"
            },
            {
              "slug": "10-minute-stress-walk",
              "canonical_id": "brali:protocol:10-minute-stress-walk",
              "action": "When stress is building, consider a short, comfortable walk if walking is safe and practical for you. Treat ten minutes as a convenient boundary, not a medically proven dose. If a greener route is just as easy and safe, prefer it as a reasonable experiment rather than making a special trip. Then notice whether you feel any different before returning to the next task.",
              "evidence_state": "reviewed",
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "exercise-snacks-real-world-t2d-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "In this specific real-world type 2 diabetes sample, the four-by-one-minute vigorous exercise-snack program was feasible and improved 30-second sit-to-stand performance more than an active mobility/stretching comparator. The trial does not show a broad cardiometabolic or aerobic-fitness advantage for this exact prescription.",
              "limitations": [
                "The sample was limited to insufficiently active adults with non-insulin-treated type 2 diabetes and well-controlled glycemia at baseline.",
                "The primary outcome was feasibility; efficacy outcomes were secondary and the study was not a definitive comparative-effectiveness trial for all health outcomes.",
                "The mobility/stretching control was active rather than no-exercise, so the contrast estimates the added effect of the vigorous condition over a matched low-intensity routine.",
                "The intervention was remotely delivered and real-world exercise intensity was lower than some prior supervised laboratory exercise-snack protocols.",
                "Estimated VO2max relied on a submaximal predictive test that the investigators reported performed poorly in this population, limiting interpretation of that fitness outcome.",
                "The trial lasted 12 weeks and does not establish longer-term adherence or clinical outcomes."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42502209/"
            },
            {
              "id": "exercise-snacks-cardiorespiratory-fitness-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "Across the included randomized studies, exercise snacks improved VO2max and peak power output on average with moderate-certainty evidence. This supports offering distributed brief exercise bouts as a practical cardiorespiratory-fitness option when a longer continuous workout is difficult to fit. The evidence does not establish one modality, timer, number of bouts, weekly frequency or intervention duration as the optimal prescription.",
              "limitations": [
                "Cardiorespiratory-fitness effects showed substantial between-study heterogeneity (I2 74% for VO2max and 63% for peak power output), which contributed to GRADE downgrading.",
                "Certainty was low for body-fat and blood-lipid outcomes because of heterogeneity, imprecision and possible publication bias for some outcomes.",
                "Interventions differed markedly in modality, intensity, bout duration, within-day distribution, weekly frequency, intervention duration and supervision.",
                "Dietary control varied across trials and could contribute to inconsistent body-composition and lipid results.",
                "Age analyses used study-level mean age rather than individual participant data, creating ecological-bias risk and making individual age cutoffs inappropriate.",
                "Potential publication bias was detected for VO2max and body-fat percentage.",
                "Subgroup analyses compared relatively small numbers of studies and cannot identify a causal optimal dose.",
                "Long-term scalability remains uncertain."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13486332/"
            },
            {
              "id": "green-route-stress-walk-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "For an existing Brali stress-walk protocol, a safe greener route can be offered as a low-friction optional modifier when it is as convenient as an ordinary route. In the reviewed randomized evidence, green exercise improved average wellbeing and affect relative to pooled non-exercise, indoor-exercise and built-up outdoor comparators. Stress-specific pooled effects favored green exercise in all three comparator categories. This supports 'prefer green when equally practical', not a stronger prescription.",
              "limitations": [
                "Several pooled outcomes showed substantial or high heterogeneity, although sensitivity analyses suggested some results were driven by a small number of studies.",
                "Most individual studies were small, and the authors noted incomplete reporting of randomization procedures and limited blinding in some trials.",
                "Most interventions were walking, so generalization to other forms or intensities of exercise is limited.",
                "Most interventions were short-term, with follow-up generally under three months, so long-term durability is uncertain.",
                "Settings, exercise formats and psychological measures varied across studies.",
                "Stress-specific comparisons contained relatively few studies: three versus non-exercise, six versus indoor exercise and two versus built-up exercise.",
                "The meta-analysis does not isolate a unique biological or psychological mechanism for any observed advantage of greener settings."
              ],
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "strength-routine",
      "category": "movement",
      "language": "en",
      "query": "How can I build a sustainable strength and mobility routine?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "fitness-strength"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "cardio-health-daily-habits",
          "10-minute-morning-stretch-routine",
          "automate-routine-data-reports",
          "temptation-bundling",
          "10-minute-stress-walk"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "fitness-strength",
          "career",
          "fear-confidence"
        ],
        "protocol_slugs": [
          "10-minute-morning-stretch-routine",
          "4-minute-hiit-tabata-workout",
          "4-minute-tabata-hiit-timer",
          "cardio-health-daily-habits",
          "safety-statements-exposure-anxiety"
        ],
        "evidence_decision_ids": [
          "exercise-snacks-real-world-t2d-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "10-minute-morning-stretch-routine",
              "canonical_id": "brali:protocol:10-minute-morning-stretch-routine",
              "action": "Use a short, gentle flexibility routine as a morning movement cue if that timing suits you. Move slowly only as far as is comfortable, start with a few simple movements, and build up gradually rather than forcing a fixed 10-to-20-minute progression.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/live-well/exercise/flexibility-exercises/"
            },
            {
              "slug": "4-minute-hiit-tabata-workout",
              "canonical_id": "brali:protocol:4-minute-hiit-tabata-workout",
              "action": "If vigorous exercise is appropriate for your current fitness and health, choose a simple movement you can perform safely. You can use it as one short interval session, or as brief exercise snacks separated across the day when a full workout is unlikely. Make the effort challenging but controlled. Treat 20/10, 3 × 1 minute, 4 × 1 minute and similar timers as formats to adapt, not scientifically optimized doses.",
              "evidence_state": "reviewed",
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC13486332/"
            },
            {
              "slug": "4-minute-tabata-hiit-timer",
              "canonical_id": "brali:protocol:4-minute-tabata-hiit-timer",
              "action": "If very vigorous exercise is suitable for you, use an interval timer to alternate a short period of harder work with recovery. A 20-second work and 10-second recovery pattern can be one option, but choose a movement and number of rounds you can control safely rather than treating 8 all-out rounds as mandatory.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/live-well/exercise/physical-activity-guidelines-for-adults-aged-19-to-64/"
            },
            {
              "slug": "cardio-health-daily-habits",
              "canonical_id": "brali:protocol:cardio-health-daily-habits",
              "action": "Look at the coming week, mark the activity you already do, then add realistic walking, moderate or vigorous activity, and strength sessions where they fit. Build gradually rather than treating the guideline as a one-day target.",
              "evidence_state": "reviewed",
              "source_url": "https://www.heart.org/en/healthy-living/healthy-lifestyle/lifes-essential-8/how-to-be-more-active-fact-sheet"
            },
            {
              "slug": "safety-statements-exposure-anxiety",
              "canonical_id": "brali:protocol:safety-statements-exposure-anxiety",
              "action": "Choose one situation you avoid because it feels frightening but that is not actually dangerous. Break it into smaller steps, start with a manageable step, approach it gradually, and notice what happens rather than trying to force fear away with a required phrase or ritual.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nhs.uk/every-mind-matters/mental-wellbeing-tips/self-help-cbt-techniques/facing-your-fears/"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "exercise-snacks-real-world-t2d-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "In this specific real-world type 2 diabetes sample, the four-by-one-minute vigorous exercise-snack program was feasible and improved 30-second sit-to-stand performance more than an active mobility/stretching comparator. The trial does not show a broad cardiometabolic or aerobic-fitness advantage for this exact prescription.",
              "limitations": [
                "The sample was limited to insufficiently active adults with non-insulin-treated type 2 diabetes and well-controlled glycemia at baseline.",
                "The primary outcome was feasibility; efficacy outcomes were secondary and the study was not a definitive comparative-effectiveness trial for all health outcomes.",
                "The mobility/stretching control was active rather than no-exercise, so the contrast estimates the added effect of the vigorous condition over a matched low-intensity routine.",
                "The intervention was remotely delivered and real-world exercise intensity was lower than some prior supervised laboratory exercise-snack protocols.",
                "Estimated VO2max relied on a submaximal predictive test that the investigators reported performed poorly in this population, limiting interpretation of that fitness outcome.",
                "The trial lasted 12 weeks and does not establish longer-term adherence or clinical outcomes."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42502209/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "recovery-breaks",
      "category": "movement",
      "language": "en",
      "query": "How should I think about breaks and recovery when my energy drops?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "recovery-energy"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "20-20-20-eye-break-reminder",
          "25-minute-pomodoro-focus-sprints",
          "30-min-eye-break-tracker",
          "90-30-focus-rest-schedule",
          "ad-astra-per-aspera-motivation-tracker"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "recovery-energy",
          "movement",
          "critical-thinking"
        ],
        "protocol_slugs": [
          "20-20-20-eye-break-reminder",
          "90-30-focus-rest-schedule",
          "temptation-bundling",
          "10-minute-stress-walk",
          "cardio-health-daily-habits"
        ],
        "evidence_decision_ids": [
          "temptation-bundling-exercise-boundary-2020",
          "green-route-stress-walk-boundary-2026",
          "genai-learning-augmentation-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "20-20-20-eye-break-reminder",
              "canonical_id": "brali:protocol:20-20-20-eye-break-reminder",
              "action": "During long periods of screen use, take a short visual break about every 20 minutes and look at something about 20 feet away for 20 seconds. Use the reminder as a simple screen-rest cue, not as a treatment or a substitute for eye care.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nei.nih.gov/eye-health-information/healthy-vision/how-eyes-work/keep-your-eyes-healthy"
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "slug": "10-minute-stress-walk",
              "canonical_id": "brali:protocol:10-minute-stress-walk",
              "action": "When stress is building, consider a short, comfortable walk if walking is safe and practical for you. Treat ten minutes as a convenient boundary, not a medically proven dose. If a greener route is just as easy and safe, prefer it as a reasonable experiment rather than making a special trip. Then notice whether you feel any different before returning to the next task.",
              "evidence_state": "reviewed",
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "slug": "cardio-health-daily-habits",
              "canonical_id": "brali:protocol:cardio-health-daily-habits",
              "action": "Look at the coming week, mark the activity you already do, then add realistic walking, moderate or vigorous activity, and strength sessions where they fit. Build gradually rather than treating the guideline as a one-day target.",
              "evidence_state": "reviewed",
              "source_url": "https://www.heart.org/en/healthy-living/healthy-lifestyle/lifes-essential-8/how-to-be-more-active-fact-sheet"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "id": "green-route-stress-walk-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "For an existing Brali stress-walk protocol, a safe greener route can be offered as a low-friction optional modifier when it is as convenient as an ordinary route. In the reviewed randomized evidence, green exercise improved average wellbeing and affect relative to pooled non-exercise, indoor-exercise and built-up outdoor comparators. Stress-specific pooled effects favored green exercise in all three comparator categories. This supports 'prefer green when equally practical', not a stronger prescription.",
              "limitations": [
                "Several pooled outcomes showed substantial or high heterogeneity, although sensitivity analyses suggested some results were driven by a small number of studies.",
                "Most individual studies were small, and the authors noted incomplete reporting of randomization procedures and limited blinding in some trials.",
                "Most interventions were walking, so generalization to other forms or intensities of exercise is limited.",
                "Most interventions were short-term, with follow-up generally under three months, so long-term durability is uncertain.",
                "Settings, exercise formats and psychological measures varied across studies.",
                "Stress-specific comparisons contained relatively few studies: three versus non-exercise, six versus indoor exercise and two versus built-up exercise.",
                "The meta-analysis does not isolate a unique biological or psychological mechanism for any observed advantage of greener settings."
              ],
              "source_url": "https://www.frontiersin.org/journals/psychology/articles/10.3389/fpsyg.2026.1802759/full"
            },
            {
              "id": "genai-learning-augmentation-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "Brali should not treat generative AI as a learning intervention by itself. In this highly heterogeneous literature, the apparent positive pooled effect did not survive robust publication-bias correction, while more informative moderator evidence suggested that outcomes depend partly on what cognitive work the learner still performs. This supports the existing human-first AI collaboration principle in learning contexts: preserve a meaningful learner attempt, explanation, retrieval, reasoning or verification step and use AI to extend or refine that activity rather than silently replacing the activity being learned.",
              "limitations": [
                "Between-study heterogeneity was extreme (I²=96.32%), and the prediction interval included negative, null and strongly positive true effects.",
                "Funnel-plot asymmetry and the Robust Bayesian Meta-Analysis indicated substantial publication bias; after correction the evidence favored no stable overall positive or negative main effect.",
                "The review's risk-of-bias assessment found high risk across included studies, with no study meeting the low-risk criteria.",
                "Many studies used cognitively incomparable intervention and control conditions; the authors identified numerous apparently large effects where this comparison problem was present.",
                "Most studies used text-based systems and many were conducted in higher education, limiting generalization to other learners, modalities and tasks.",
                "AI literacy, metacognitive skill, delegation behavior, prompt quality and verification behavior were often underreported, leaving important mechanisms unresolved.",
                "Evidence about learner challenges and instructional supports was sparse and inconsistently reported, so those qualitative findings should not be turned into general effect estimates.",
                "Moderator patterns explained only part of the heterogeneity and did not produce a sufficient if-then configuration that guaranteed large effects."
              ],
              "source_url": "https://link.springer.com/article/10.1007/s10462-026-11665-9"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "memory-study",
      "category": "memory",
      "language": "en",
      "query": "How can I remember what I study instead of only rereading it?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "memory"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "active-recall-test-yourself",
          "spaced-recall-coach",
          "10-minute-language-microsprints",
          "10-minute-morning-stretch-routine",
          "2-minute-desk-tidy-timer"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "memory"
        ],
        "protocol_slugs": [
          "active-recall-test-yourself",
          "spaced-recall-coach",
          "avoid-list-interference-memory-retention",
          "prequestions-before-learning",
          "self-explain-what-you-learn"
        ],
        "evidence_decision_ids": [
          "retrieval-procedural-application-2026",
          "testing-effect-direct-forward-2026",
          "testing-versus-restudy-retention-boundary-2014"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "spaced-recall-coach",
              "canonical_id": "brali:protocol:spaced-recall-coach",
              "action": "Choose one set of verbal material and decide how long you need to retain it. Split the planned repetitions across at least two sessions separated by time. Use a longer gap when the required retention period is longer, then check final recall and adjust the next schedule instead of treating one interval as universally optimal.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0033-2909.132.3.354"
            },
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            },
            {
              "slug": "self-explain-what-you-learn",
              "canonical_id": "brali:protocol:self-explain-what-you-learn",
              "action": "Study a manageable chunk, then look away from the explanation and produce a brief self-explanation: what the idea means, why a step follows, how it connects to what you already know, or when it would apply. Compare your explanation with the source, identify one gap or error, and revise it. Use the prompts that fit the material; do not turn every paragraph into a compulsory monologue.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10001-x"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "retrieval-procedural-application-2026",
              "decision": "challenge-existing",
              "supported_claim": "Adding retrieval practice to worked examples improved long-term retention of the spelling rules in this classroom study, but the combined group did not outperform worked examples alone on correct rule application after one week.",
              "limitations": [
                "The sample was 105 fourth-grade pupils learning Dutch verb spelling in a small number of intact classes and schools.",
                "The comparison tested retrieval practice added to worked examples, not retrieval practice in isolation.",
                "The target was one form of procedural knowledge and the study does not establish identical boundaries in other domains or age groups.",
                "The result supports separating retention from application outcomes, not concluding that retrieval practice is ineffective for complex learning generally."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/pii/S0959475226000629"
            },
            {
              "id": "testing-effect-direct-forward-2026",
              "decision": "challenge-existing",
              "supported_claim": "Retrieval practice can improve memory, but the literature commonly labeled as the testing effect mixes at least two distinct effects. In this meta-analysis, 66% of included testing effects came from mixed studies and 34% from direct studies, with mixed studies producing larger effects.",
              "limitations": [
                "The meta-analysis focuses on free-recall testing-effect studies rather than every form of retrieval practice or every educational outcome.",
                "Its main contribution is separation of direct and forward effects; it does not negate the broader evidence that retrieval practice can support retention.",
                "The reported mixture of study designs means older aggregate claims may need narrower interpretation rather than wholesale rejection."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42258276/"
            },
            {
              "id": "testing-versus-restudy-retention-boundary-2014",
              "decision": "support-existing",
              "supported_claim": "Across the included testing-versus-restudy literature, retrieval testing produced better later retention on average than equivalent-duration restudy. The pooled random-effects estimate was positive, but effects varied substantially across studies and testing conditions. This supports Brali's bounded recommendation to retrieve information from memory, check the answer, and use that loop when retention is the target.",
              "limitations": [
                "Between-study heterogeneity in the primary analysis was very high, so the pooled mean does not describe one uniform effect across designs or contexts.",
                "The review deliberately restricted the quantitative synthesis to testing-versus-equivalent-restudy contrasts and does not represent every retrieval-practice paradigm.",
                "Testing benefits varied with methodological factors including initial test format, retention interval and feedback.",
                "Many contributing studies used controlled experimental learning tasks and college samples, which limits direct extrapolation to every real-world learning context.",
                "The review's main outcome is retention; it does not establish that remembered information can be applied correctly in a novel task or procedure.",
                "Theoretical findings did not provide one complete cohesive mechanism for all testing phenomena, so Brali should avoid mechanism decoration.",
                "A positive average effect does not imply that every individual effect was positive; the reviewed distribution included negative and null estimates."
              ],
              "source_url": "https://doi.org/10.1037/a0037559"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "wakeful-rest-boundary",
      "category": "memory",
      "language": "en",
      "query": "Is ten minutes of quiet rest after learning proven to improve memory for everyone?",
      "mode": "bounded-evidence",
      "expected": {
        "topic_ids": [
          "memory"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "avoid-list-interference-memory-retention"
        ],
        "evidence_decision_ids": [
          "wakeful-rest-memory-2026"
        ]
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "avoid-list-interference-memory-retention",
          "10-minute-stress-walk",
          "10-minute-language-microsprints",
          "20-20-20-eye-break-reminder",
          "4-minute-tabata-hiit-timer"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "memory",
          "language-learning",
          "skill-learning"
        ],
        "protocol_slugs": [
          "avoid-list-interference-memory-retention",
          "active-recall-test-yourself",
          "ai-learning-attempt-feedback-retest",
          "prequestions-before-learning",
          "self-explain-what-you-learn"
        ],
        "evidence_decision_ids": [
          "wakeful-rest-memory-2026",
          "wakeful-rest-complex-learning-challenge-2026",
          "testing-effect-direct-forward-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": 1,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            },
            {
              "slug": "self-explain-what-you-learn",
              "canonical_id": "brali:protocol:self-explain-what-you-learn",
              "action": "Study a manageable chunk, then look away from the explanation and produce a brief self-explanation: what the idea means, why a step follows, how it connects to what you already know, or when it would apply. Compare your explanation with the source, identify one gap or error, and revise it. Use the prompts that fit the material; do not turn every paragraph into a compulsory monologue.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10001-x"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "wakeful-rest-memory-2026",
              "decision": "propose-protocol",
              "supported_claim": "A short period of quiet, minimally interfering wakeful rest after learning can reduce forgetting on average. The pooled effect is substantially smaller and less consistent in healthy young to middle-aged adults than in older adults or patient samples.",
              "limitations": [
                "There was moderate to large between-study heterogeneity overall.",
                "The overall prediction interval included negative effects, so future studies are not guaranteed to find a benefit.",
                "Healthy young to middle-aged adults had the smallest pooled effect (g=0.20 in the authors' group analysis).",
                "Patient studies were small and showed funnel-plot asymmetry consistent with possible publication bias.",
                "Learning materials, interference tasks, test procedures, age, and clinical status varied across studies.",
                "The meta-analysis concerns declarative memory and does not justify broad claims about skill acquisition, creativity, focus, or general intelligence."
              ],
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "id": "wakeful-rest-complex-learning-challenge-2026",
              "decision": "challenge-existing",
              "supported_claim": "The existing Brali wakeful-rest protocol must not promise better comprehension or long-term retention for complex educational reading. In this more ecologically valid randomized study, eight minutes of post-reading wakeful rest did not produce a consistent advantage over social media, math or similar-text reading on immediate or one-week factual and conceptual tests. Keep the protocol narrow: it is an optional declarative-memory tactic, not a general learning upgrade.",
              "limitations": [
                "The study used one expository learning task in a university-student population.",
                "The wakeful-rest condition used an autogenic-training implementation and therefore does not represent every possible quiet-rest procedure.",
                "Only one eight-minute post-learning duration was tested.",
                "The delayed test used different but conceptually aligned questions rather than identical items from the immediate test.",
                "A single null study cannot establish absence of a wakeful-rest effect across all complex learning tasks or populations."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12945945/"
            },
            {
              "id": "testing-effect-direct-forward-2026",
              "decision": "challenge-existing",
              "supported_claim": "Retrieval practice can improve memory, but the literature commonly labeled as the testing effect mixes at least two distinct effects. In this meta-analysis, 66% of included testing effects came from mixed studies and 34% from direct studies, with mixed studies producing larger effects.",
              "limitations": [
                "The meta-analysis focuses on free-recall testing-effect studies rather than every form of retrieval practice or every educational outcome.",
                "Its main contribution is separation of direct and forward effects; it does not negate the broader evidence that retrieval practice can support retention.",
                "The reported mixture of study designs means older aggregate claims may need narrower interpretation rather than wholesale rejection."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42258276/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "active-recall-exact",
      "category": "learning",
      "language": "en",
      "query": "Show me an active recall protocol where I test myself",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "memory",
          "skill-learning"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "active-recall-test-yourself"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "avoid-list-interference-memory-retention",
          "active-recall-test-yourself",
          "ai-test-case-generator",
          "consider-the-opposite",
          "create-win-win-outcomes"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "memory",
          "conflict-repair",
          "problem-solving"
        ],
        "protocol_slugs": [
          "avoid-list-interference-memory-retention",
          "active-recall-test-yourself",
          "spaced-recall-coach",
          "ai-learning-attempt-feedback-retest",
          "prequestions-before-learning"
        ],
        "evidence_decision_ids": [
          "wakeful-rest-complex-learning-challenge-2026",
          "wakeful-rest-memory-2026",
          "retrieval-procedural-application-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "spaced-recall-coach",
              "canonical_id": "brali:protocol:spaced-recall-coach",
              "action": "Choose one set of verbal material and decide how long you need to retain it. Split the planned repetitions across at least two sessions separated by time. Use a longer gap when the required retention period is longer, then check final recall and adjust the next schedule instead of treating one interval as universally optimal.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0033-2909.132.3.354"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "wakeful-rest-complex-learning-challenge-2026",
              "decision": "challenge-existing",
              "supported_claim": "The existing Brali wakeful-rest protocol must not promise better comprehension or long-term retention for complex educational reading. In this more ecologically valid randomized study, eight minutes of post-reading wakeful rest did not produce a consistent advantage over social media, math or similar-text reading on immediate or one-week factual and conceptual tests. Keep the protocol narrow: it is an optional declarative-memory tactic, not a general learning upgrade.",
              "limitations": [
                "The study used one expository learning task in a university-student population.",
                "The wakeful-rest condition used an autogenic-training implementation and therefore does not represent every possible quiet-rest procedure.",
                "Only one eight-minute post-learning duration was tested.",
                "The delayed test used different but conceptually aligned questions rather than identical items from the immediate test.",
                "A single null study cannot establish absence of a wakeful-rest effect across all complex learning tasks or populations."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12945945/"
            },
            {
              "id": "wakeful-rest-memory-2026",
              "decision": "propose-protocol",
              "supported_claim": "A short period of quiet, minimally interfering wakeful rest after learning can reduce forgetting on average. The pooled effect is substantially smaller and less consistent in healthy young to middle-aged adults than in older adults or patient samples.",
              "limitations": [
                "There was moderate to large between-study heterogeneity overall.",
                "The overall prediction interval included negative effects, so future studies are not guaranteed to find a benefit.",
                "Healthy young to middle-aged adults had the smallest pooled effect (g=0.20 in the authors' group analysis).",
                "Patient studies were small and showed funnel-plot asymmetry consistent with possible publication bias.",
                "Learning materials, interference tasks, test procedures, age, and clinical status varied across studies.",
                "The meta-analysis concerns declarative memory and does not justify broad claims about skill acquisition, creativity, focus, or general intelligence."
              ],
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "id": "retrieval-procedural-application-2026",
              "decision": "challenge-existing",
              "supported_claim": "Adding retrieval practice to worked examples improved long-term retention of the spelling rules in this classroom study, but the combined group did not outperform worked examples alone on correct rule application after one week.",
              "limitations": [
                "The sample was 105 fourth-grade pupils learning Dutch verb spelling in a small number of intact classes and schools.",
                "The comparison tested retrieval practice added to worked examples, not retrieval practice in isolation.",
                "The target was one form of procedural knowledge and the study does not establish identical boundaries in other domains or age groups.",
                "The result supports separating retention from application outcomes, not concluding that retrieval practice is ineffective for complex learning generally."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/pii/S0959475226000629"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "skill-feedback",
      "category": "learning",
      "language": "en",
      "query": "How can I learn a skill through practice and feedback instead of passive reading?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "skill-learning"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "weekly-theme-learning-sprints",
          "10-minute-language-microsprints",
          "ai-learning-attempt-feedback-retest",
          "prequestions-before-learning",
          "active-recall-test-yourself"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "skill-learning",
          "language-learning",
          "revision-feedback"
        ],
        "protocol_slugs": [
          "contextual-language-coach",
          "weekly-theme-learning-sprints",
          "ai-learning-attempt-feedback-retest",
          "active-recall-test-yourself",
          "prequestions-before-learning"
        ],
        "evidence_decision_ids": [
          "testing-versus-restudy-retention-boundary-2014",
          "retrieval-procedural-application-2026",
          "testing-effect-direct-forward-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "contextual-language-coach",
              "canonical_id": "brali:protocol:contextual-language-coach",
              "action": "Define the scenario and desired outcome, write the core request or response, add likely follow-up phrases and one repair phrase for misunderstanding, rehearse both sides aloud, use the interaction when appropriate, and update the phrase set from the point where you hesitated or needed clarification.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "weekly-theme-learning-sprints",
              "canonical_id": "brali:protocol:weekly-theme-learning-sprints",
              "action": "Choose one narrow learning theme for the week, define one visible result for Friday, and complete at least one small practice rep on each day you work on it.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "testing-versus-restudy-retention-boundary-2014",
              "decision": "support-existing",
              "supported_claim": "Across the included testing-versus-restudy literature, retrieval testing produced better later retention on average than equivalent-duration restudy. The pooled random-effects estimate was positive, but effects varied substantially across studies and testing conditions. This supports Brali's bounded recommendation to retrieve information from memory, check the answer, and use that loop when retention is the target.",
              "limitations": [
                "Between-study heterogeneity in the primary analysis was very high, so the pooled mean does not describe one uniform effect across designs or contexts.",
                "The review deliberately restricted the quantitative synthesis to testing-versus-equivalent-restudy contrasts and does not represent every retrieval-practice paradigm.",
                "Testing benefits varied with methodological factors including initial test format, retention interval and feedback.",
                "Many contributing studies used controlled experimental learning tasks and college samples, which limits direct extrapolation to every real-world learning context.",
                "The review's main outcome is retention; it does not establish that remembered information can be applied correctly in a novel task or procedure.",
                "Theoretical findings did not provide one complete cohesive mechanism for all testing phenomena, so Brali should avoid mechanism decoration.",
                "A positive average effect does not imply that every individual effect was positive; the reviewed distribution included negative and null estimates."
              ],
              "source_url": "https://doi.org/10.1037/a0037559"
            },
            {
              "id": "retrieval-procedural-application-2026",
              "decision": "challenge-existing",
              "supported_claim": "Adding retrieval practice to worked examples improved long-term retention of the spelling rules in this classroom study, but the combined group did not outperform worked examples alone on correct rule application after one week.",
              "limitations": [
                "The sample was 105 fourth-grade pupils learning Dutch verb spelling in a small number of intact classes and schools.",
                "The comparison tested retrieval practice added to worked examples, not retrieval practice in isolation.",
                "The target was one form of procedural knowledge and the study does not establish identical boundaries in other domains or age groups.",
                "The result supports separating retention from application outcomes, not concluding that retrieval practice is ineffective for complex learning generally."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/pii/S0959475226000629"
            },
            {
              "id": "testing-effect-direct-forward-2026",
              "decision": "challenge-existing",
              "supported_claim": "Retrieval practice can improve memory, but the literature commonly labeled as the testing effect mixes at least two distinct effects. In this meta-analysis, 66% of included testing effects came from mixed studies and 34% from direct studies, with mixed studies producing larger effects.",
              "limitations": [
                "The meta-analysis focuses on free-recall testing-effect studies rather than every form of retrieval practice or every educational outcome.",
                "Its main contribution is separation of direct and forward effects; it does not negate the broader evidence that retrieval practice can support retention.",
                "The reported mixture of study designs means older aggregate claims may need narrower interpretation rather than wholesale rejection."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42258276/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "language-vocabulary",
      "category": "learning",
      "language": "en",
      "query": "How can I practice vocabulary and useful phrases in a new language?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "language-learning"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "10-minute-language-microsprints",
          "contextual-language-coach",
          "savor-positive-moment",
          "5-minute-meditation-habit-tracker",
          "active-recall-test-yourself"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "language-learning",
          "attention-focus",
          "environment-design"
        ],
        "protocol_slugs": [
          "10-minute-language-microsprints",
          "contextual-language-coach",
          "batch-non-urgent-notifications",
          "refocus-present-stop-mind-wandering",
          "ready-to-resume-plan"
        ],
        "evidence_decision_ids": [
          "ready-to-resume-interruption-boundary-2018",
          "vocabulary-pretesting-guess-feedback-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 2,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "10-minute-language-microsprints",
              "canonical_id": "brali:protocol:10-minute-language-microsprints",
              "action": "Pick one narrow language task, such as reviewing a few phrases, listening to a short clip, reading a small passage, or recording a brief spoken answer. Practice for a short session and finish with one note about what to revisit next.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "contextual-language-coach",
              "canonical_id": "brali:protocol:contextual-language-coach",
              "action": "Define the scenario and desired outcome, write the core request or response, add likely follow-up phrases and one repair phrase for misunderstanding, rehearse both sides aloud, use the interaction when appropriate, and update the phrase set from the point where you hesitated or needed clarification.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "ready-to-resume-interruption-boundary-2018",
              "decision": "propose-protocol",
              "supported_claim": "When unfinished work must be interrupted, a short resumption plan can reduce attention residue and protect performance in the studied interruption contexts.",
              "limitations": [
                "The evidence comes from a small set of controlled interruption studies and does not represent every form of complex, collaborative or high-stakes real-world work.",
                "A ready-to-resume plan can mitigate attention residue in the studied contexts; it does not make task switching cost-free or imply that avoidable interruptions should be accepted.",
                "The research supports making a concrete resumption plan, not Brali's exact three-prompt note format, note length, writing medium or timing rule.",
                "The studies do not establish a universal productivity percentage, a guaranteed performance benefit, or the same effect for every individual and task."
              ],
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "id": "vocabulary-pretesting-guess-feedback-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For adults learning new, concrete second-language vocabulary paired with images, adding a forced multiple-choice guess before immediately revealing the correct pairing can produce a modest short-delay memory advantage over simply reading the pair. A bounded Brali protocol may therefore use a cue → guess → immediate correct feedback → later recall sequence for vocabulary practice, while keeping the claim limited to this learning format and time horizon.",
              "limitations": [
                "All criterial tests followed a short same-session distractor period rather than a delayed retention interval, so long-term retention was not established.",
                "The studies used concrete Spanish nouns paired with images; vocabulary type and language-learning context were narrow.",
                "Participants were adult Prolific users from English-speaking countries and exclusions were substantial in several experiments.",
                "Pretesting trials included up to 8 seconds of cue-only guessing before the same 5 seconds of correct pair exposure used in reading, so total cue exposure was longer in the pretesting condition.",
                "Initial guess accuracy was about 35–38%, raising the possibility of some prior familiarity despite exclusions; the authors note that benefits also appeared for incorrectly guessed items.",
                "Learning condition was primarily within-subjects, and the authors identify between-subject replication as a future research need.",
                "Multiple-choice benefits were not significant in Experiment 3, so recognition effects were less uniform than cued-recall effects."
              ],
              "source_url": "https://link.springer.com/article/10.1186/s41235-026-00708-y"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "weekly-theme-learning",
      "category": "learning",
      "language": "en",
      "query": "How can I spend each week focused on a single learning theme?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [],
        "acceptable_topic_ids": [
          "skill-learning"
        ],
        "protocol_slugs": [
          "weekly-theme-learning-sprints"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "weekly-theme-learning-sprints",
          "avoid-list-interference-memory-retention",
          "3-minute-sensory-mindfulness-act",
          "ask-a-peer-to-proofread",
          "ideal-sleep-hours-finder"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "attention-focus",
          "language-learning",
          "skill-learning"
        ],
        "protocol_slugs": [
          "weekly-theme-learning-sprints",
          "90-30-focus-rest-schedule",
          "contextual-language-coach",
          "batch-non-urgent-notifications",
          "ready-to-resume-plan"
        ],
        "evidence_decision_ids": [
          "notification-batching-boundary-2019",
          "testing-effect-direct-forward-2026",
          "vocabulary-pretesting-guess-feedback-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "weekly-theme-learning-sprints",
              "canonical_id": "brali:protocol:weekly-theme-learning-sprints",
              "action": "Choose one narrow learning theme for the week, define one visible result for Friday, and complete at least one small practice rep on each day you work on it.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "contextual-language-coach",
              "canonical_id": "brali:protocol:contextual-language-coach",
              "action": "Define the scenario and desired outcome, write the core request or response, add likely follow-up phrases and one repair phrase for misunderstanding, rehearse both sides aloud, use the interaction when appropriate, and update the phrase set from the point where you hesitated or needed clarification.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "notification-batching-boundary-2019",
              "decision": "propose-protocol",
              "supported_claim": "Predictable batching of non-urgent notifications is a defensible attention-environment experiment. In this field trial, three daily batches reduced perceived interruption and stress relative to usual delivery, while hourly batching changed little and complete notification removal increased anxiety/FoMO. Brali should preserve urgent exceptions and treat schedule frequency as a user-fit parameter rather than a fixed dose.",
              "limitations": [
                "Single short field experiment rather than a replicated long-term evidence base.",
                "Participants were recruited through an online labor market and may not represent all smartphone users or work contexts.",
                "The intervention used a custom notification-management implementation.",
                "Many psychological outcomes were self-reported and the study intentionally used broad exploratory measurement.",
                "The study compared a small set of schedules and cannot identify an optimal individualized batching frequency."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "id": "testing-effect-direct-forward-2026",
              "decision": "challenge-existing",
              "supported_claim": "Retrieval practice can improve memory, but the literature commonly labeled as the testing effect mixes at least two distinct effects. In this meta-analysis, 66% of included testing effects came from mixed studies and 34% from direct studies, with mixed studies producing larger effects.",
              "limitations": [
                "The meta-analysis focuses on free-recall testing-effect studies rather than every form of retrieval practice or every educational outcome.",
                "Its main contribution is separation of direct and forward effects; it does not negate the broader evidence that retrieval practice can support retention.",
                "The reported mixture of study designs means older aggregate claims may need narrower interpretation rather than wholesale rejection."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42258276/"
            },
            {
              "id": "vocabulary-pretesting-guess-feedback-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For adults learning new, concrete second-language vocabulary paired with images, adding a forced multiple-choice guess before immediately revealing the correct pairing can produce a modest short-delay memory advantage over simply reading the pair. A bounded Brali protocol may therefore use a cue → guess → immediate correct feedback → later recall sequence for vocabulary practice, while keeping the claim limited to this learning format and time horizon.",
              "limitations": [
                "All criterial tests followed a short same-session distractor period rather than a delayed retention interval, so long-term retention was not established.",
                "The studies used concrete Spanish nouns paired with images; vocabulary type and language-learning context were narrow.",
                "Participants were adult Prolific users from English-speaking countries and exclusions were substantial in several experiments.",
                "Pretesting trials included up to 8 seconds of cue-only guessing before the same 5 seconds of correct pair exposure used in reading, so total cue exposure was longer in the pretesting condition.",
                "Initial guess accuracy was about 35–38%, raising the possibility of some prior familiarity despite exclusions; the authors note that benefits also appeared for incorrectly guessed items.",
                "Learning condition was primarily within-subjects, and the authors identify between-subject replication as a future research need.",
                "Multiple-choice benefits were not significant in Experiment 3, so recognition effects were less uniform than cued-recall effects."
              ],
              "source_url": "https://link.springer.com/article/10.1186/s41235-026-00708-y"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "critical-claim",
      "category": "thinking",
      "language": "en",
      "query": "How can I check a claim against evidence before believing it?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "critical-thinking"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "prequestions-before-learning",
          "ai-learning-attempt-feedback-retest",
          "ai-customer-reply-coach",
          "ai-decision-alternatives",
          "ai-idea-divergence-pass"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "cognitive-biases",
          "critical-thinking",
          "metacognition"
        ],
        "protocol_slugs": [
          "consider-the-opposite",
          "assumed-similarity-bias-check",
          "active-listening-meeting-notes",
          "anticipate-sudden-trend-shifts",
          "prequestions-before-learning"
        ],
        "evidence_decision_ids": [
          "consider-opposite-social-judgment-boundary-1984",
          "debiasing-education-transfer-boundary-2025",
          "prequestions-targeted-learning-boundary-2025"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "consider-the-opposite",
              "canonical_id": "brali:protocol:consider-the-opposite",
              "action": "Write your current conclusion in one sentence. Then ask: what evidence, mechanism, or alternative explanation could make the opposite conclusion reasonable? Generate a concrete alternative rather than telling yourself to be objective. Check whether your decision would change if that alternative were true. Use this on decisions where biased assimilation or one-sided hypothesis testing is a real risk; do not turn it into endless doubt about routine choices.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "slug": "assumed-similarity-bias-check",
              "canonical_id": "brali:protocol:assumed-similarity-bias-check",
              "action": "When interacting with others: - Ask: \"What’s unique about this person’s perspective or experience?\" - Listen actively: Avoid projecting your own traits or beliefs onto them. - Reflect: Notice when you assume similarities without evidence and correct yourself. Example: Don’t assume a coworker has the same work style as you; ask about their preferences and needs.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "active-listening-meeting-notes",
              "canonical_id": "brali:protocol:active-listening-meeting-notes",
              "action": "To stay sharp: - Take notes: Write down key points from the person speaking before you. - Breathe and listen: Avoid rehearsing your own response while someone else is speaking. - Repeat mentally: After someone speaks, quickly repeat their main point in your head. Example: In a team meeting, note what the person before you says and reference it when it’s your turn.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "anticipate-sudden-trend-shifts",
              "canonical_id": "brali:protocol:anticipate-sudden-trend-shifts",
              "action": "When predicting trends: - Ask yourself: \"Could there be sudden changes or breaks in this trend?\" - Prepare for shifts: Always consider outliers and unexpected events in planning. - Look for weak signals: Early signs of change can help you adapt quickly. Example: A stock steadily rising? Plan for a downturn just as much as for continued growth.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "consider-opposite-social-judgment-boundary-1984",
              "decision": "propose-protocol",
              "supported_claim": "When a judgment is vulnerable to one-sided evidence processing, deliberately generating an opposed possibility can reduce bias on some tasks more effectively than simply telling oneself to be fair or unbiased. Brali can use this as a concrete pre-decision check while preserving the possibility that the original conclusion remains correct.",
              "limitations": [
                "Classic laboratory/social-judgment evidence from undergraduate samples.",
                "Only two focal task domains were tested in the original article.",
                "Long-term persistence and broad transfer were not established.",
                "The authors note that considering the opposite can in some circumstances overweight disconfirming evidence and create a different form of partiality.",
                "Demand characteristics and task-specific effects remain possible."
              ],
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "id": "debiasing-education-transfer-boundary-2025",
              "decision": "support-existing",
              "supported_claim": "Debiasing education can produce a small average improvement on targeted bias tasks. This supports modest expectations for a consider-the-opposite check while making the boundary explicit: depth of learning and transfer to meaningful real-world decisions remain uncertain.",
              "limitations": [
                "All included studies were rated unclear or high risk of bias.",
                "There was some evidence of publication bias.",
                "Interventions, bias targets and outcome measures were highly heterogeneous.",
                "Transfer beyond explicitly trained tasks was limited or uncertain in many studies.",
                "The pooled effect represents diverse educational interventions rather than the consider-the-opposite strategy alone."
              ],
              "source_url": "https://www.nature.com/articles/s41562-025-02253-y"
            },
            {
              "id": "prequestions-targeted-learning-boundary-2025",
              "decision": "propose-protocol",
              "supported_claim": "Prequestions can improve later learning of the specific information they target. The meta-analysis reported g = 0.66 for prequestioned information and g = 0.01 for non-prequestioned information; feedback alongside prequestions was associated with stronger targeted learning.",
              "limitations": [
                "The average benefit is specific to prequestioned information rather than a broad benefit to all content.",
                "The meta-analysis does not establish one universal question count, timing rule, or content format.",
                "Effect sizes summarize heterogeneous studies and do not guarantee the same result for an individual learning session."
              ],
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "problem-hypothesis",
      "category": "thinking",
      "language": "en",
      "query": "How can I frame a problem and test a hypothesis?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "problem-solving"
        ],
        "acceptable_topic_ids": [
          "strategy"
        ],
        "protocol_slugs": [
          "start-with-a-hypothesis"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "start-with-a-hypothesis",
          "biomimicry-creative-problem-solving",
          "bold-brainstorm-kickoff",
          "consider-the-opposite",
          "5-minute-meditation-habit-tracker"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "problem-solving",
          "creative-problem-solving"
        ],
        "protocol_slugs": [
          "break-down-big-problems-triz",
          "start-with-a-hypothesis",
          "ab-test-learning-loop",
          "process-of-elimination-tracker",
          "biomimicry-creative-problem-solving"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "break-down-big-problems-triz",
              "canonical_id": "brali:protocol:break-down-big-problems-triz",
              "action": "State the unwanted outcome, map the main parts and interactions, write the trade-off as two requirements that appear to conflict, generate separation or redesign options, choose a reversible check, and record which assumption or constraint changed.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "start-with-a-hypothesis",
              "canonical_id": "brali:protocol:start-with-a-hypothesis",
              "action": "When a work problem is unclear, write one plausible explanation, state what you would expect to observe if it were true, run the smallest safe check that could change your mind, and update the hypothesis from the result.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ab-test-learning-loop",
              "canonical_id": "brali:protocol:ab-test-learning-loop",
              "action": "Choose one reversible decision, define what you want to learn, compare two approaches as fairly as you can, and record the outcome before deciding what to try next. Use proper experimental design when the result needs statistical confidence.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "process-of-elimination-tracker",
              "canonical_id": "brali:protocol:process-of-elimination-tracker",
              "action": "List the viable options, write two to four constraints that are genuinely relevant to this decision, eliminate only options that clearly fail a constraint, inspect what remains for missing information and trade-offs, then choose the next reversible test or decision step.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "biomimicry-creative-problem-solving",
              "canonical_id": "brali:protocol:biomimicry-creative-problem-solving",
              "action": "State what the solution must do without naming the current solution. Choose a biological example from a reliable description. Separate the observed structure or process from the metaphor. Write one transferable principle, generate several applications and test the smallest reversible option within safety and operational constraints.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "metacognition-confidence",
      "category": "thinking",
      "language": "en",
      "query": "How can I monitor whether I really understand something instead of being overconfident?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "metacognition"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "avoid-selection-bias-in-analysis",
          "10-minute-language-microsprints",
          "10-minute-stress-walk",
          "20-20-20-eye-break-reminder",
          "25-minute-pomodoro-focus-sprints"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "metacognition",
          "clear-communication",
          "emotion-regulation"
        ],
        "protocol_slugs": [
          "refocus-present-stop-mind-wandering",
          "ai-learning-attempt-feedback-retest",
          "prequestions-before-learning",
          "self-explain-what-you-learn",
          "ask-for-feedback-tracker"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            },
            {
              "slug": "self-explain-what-you-learn",
              "canonical_id": "brali:protocol:self-explain-what-you-learn",
              "action": "Study a manageable chunk, then look away from the explanation and produce a brief self-explanation: what the idea means, why a step follows, how it connects to what you already know, or when it would apply. Compare your explanation with the source, identify one gap or error, and revise it. Use the prompts that fit the material; do not turn every paragraph into a compulsory monologue.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10001-x"
            },
            {
              "slug": "ask-for-feedback-tracker",
              "canonical_id": "brali:protocol:ask-for-feedback-tracker",
              "action": "State the intended outcome and the current decision. Ask one focused question such as what is unclear, what assumption looks wrong or what should change first. Choose a reviewer who can judge that question, then record what you will accept, test, discuss, defer or ignore.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "data-uncertainty",
      "category": "thinking",
      "language": "en",
      "query": "Show me an A B test learning loop for problem solving",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "problem-solving"
        ],
        "acceptable_topic_ids": [
          "strategy"
        ],
        "protocol_slugs": [
          "ab-test-learning-loop"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "ab-test-learning-loop",
          "ai-learning-attempt-feedback-retest",
          "avoid-list-interference-memory-retention",
          "biomimicry-creative-problem-solving",
          "bold-brainstorm-kickoff"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "problem-solving",
          "creative-problem-solving",
          "language-learning"
        ],
        "protocol_slugs": [
          "break-down-big-problems-triz",
          "ab-test-learning-loop",
          "start-with-a-hypothesis",
          "process-of-elimination-tracker",
          "biomimicry-creative-problem-solving"
        ],
        "evidence_decision_ids": [
          "testing-versus-restudy-retention-boundary-2014",
          "distributed-practice-verbal-recall-boundary-2006",
          "pmr-subjective-sleep-quality-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "break-down-big-problems-triz",
              "canonical_id": "brali:protocol:break-down-big-problems-triz",
              "action": "State the unwanted outcome, map the main parts and interactions, write the trade-off as two requirements that appear to conflict, generate separation or redesign options, choose a reversible check, and record which assumption or constraint changed.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ab-test-learning-loop",
              "canonical_id": "brali:protocol:ab-test-learning-loop",
              "action": "Choose one reversible decision, define what you want to learn, compare two approaches as fairly as you can, and record the outcome before deciding what to try next. Use proper experimental design when the result needs statistical confidence.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "start-with-a-hypothesis",
              "canonical_id": "brali:protocol:start-with-a-hypothesis",
              "action": "When a work problem is unclear, write one plausible explanation, state what you would expect to observe if it were true, run the smallest safe check that could change your mind, and update the hypothesis from the result.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "process-of-elimination-tracker",
              "canonical_id": "brali:protocol:process-of-elimination-tracker",
              "action": "List the viable options, write two to four constraints that are genuinely relevant to this decision, eliminate only options that clearly fail a constraint, inspect what remains for missing information and trade-offs, then choose the next reversible test or decision step.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "biomimicry-creative-problem-solving",
              "canonical_id": "brali:protocol:biomimicry-creative-problem-solving",
              "action": "State what the solution must do without naming the current solution. Choose a biological example from a reliable description. Separate the observed structure or process from the metaphor. Write one transferable principle, generate several applications and test the smallest reversible option within safety and operational constraints.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "testing-versus-restudy-retention-boundary-2014",
              "decision": "support-existing",
              "supported_claim": "Across the included testing-versus-restudy literature, retrieval testing produced better later retention on average than equivalent-duration restudy. The pooled random-effects estimate was positive, but effects varied substantially across studies and testing conditions. This supports Brali's bounded recommendation to retrieve information from memory, check the answer, and use that loop when retention is the target.",
              "limitations": [
                "Between-study heterogeneity in the primary analysis was very high, so the pooled mean does not describe one uniform effect across designs or contexts.",
                "The review deliberately restricted the quantitative synthesis to testing-versus-equivalent-restudy contrasts and does not represent every retrieval-practice paradigm.",
                "Testing benefits varied with methodological factors including initial test format, retention interval and feedback.",
                "Many contributing studies used controlled experimental learning tasks and college samples, which limits direct extrapolation to every real-world learning context.",
                "The review's main outcome is retention; it does not establish that remembered information can be applied correctly in a novel task or procedure.",
                "Theoretical findings did not provide one complete cohesive mechanism for all testing phenomena, so Brali should avoid mechanism decoration.",
                "A positive average effect does not imply that every individual effect was positive; the reviewed distribution included negative and null estimates."
              ],
              "source_url": "https://doi.org/10.1037/a0037559"
            },
            {
              "id": "distributed-practice-verbal-recall-boundary-2006",
              "decision": "support-existing",
              "supported_claim": "For verbal material measured by later recall, separating repeated study episodes by a meaningful interval generally supports better retention than concentrating the same material into massed study. The spacing associated with the best later recall tended to increase as the intended retention interval increased. For material that must be retained over months or years, the reviewed evidence supports distributing study across days or longer rather than completing all review in one sitting or one day. These findings justify the rewritten Brali action only within a verbal-recall boundary and without one universal schedule.",
              "limitations": [
                "The synthesis was restricted to verbal memory tasks measured by recall and deliberately excluded recognition, frequency judgments and the heterogeneous skill-learning literature.",
                "The evidence base was dominated by young adults; the authors reported very little middle-aged and older-adult evidence and insufficient long-term child data for confident generalization.",
                "Many studies did not report the variance data needed for effect-size calculation, so several analyses relied on accuracy differences and included fewer effect-size estimates.",
                "Published null findings may be underrepresented because of the file-drawer problem.",
                "Study materials, presentation schedules, retention intervals and experimental procedures varied substantially.",
                "Some historical studies confounded longer spacing with more relearning trials; the review examined this problem but could not remove every design limitation from the literature.",
                "Binning inter-study and retention intervals supported broad patterns but reduced the ability to recommend exact intervals.",
                "The useful interval depends jointly on spacing and the later retention target, and the authors stated that exact long-term optimization could not be specified with certainty.",
                "Evidence comparing expanding and fixed schedules was sparse and inconsistent, with large between-study variability.",
                "The synthesis addresses later recall, not broader comprehension, transfer, motivation, study adherence or real-world performance.",
                "The review was published in 2006 and should be rechecked against newer syntheses before adding more precise scheduling claims."
              ],
              "source_url": "https://doi.org/10.1037/0033-2909.132.3.354"
            },
            {
              "id": "pmr-subjective-sleep-quality-boundary-2026",
              "decision": "support-existing",
              "supported_claim": "Across the included randomized trials, PMR improved subjective PSQI sleep-quality scores on average in clinically heterogeneous adult populations. The direction of the pooled effect remained after sensitivity and trim-and-fill analyses. This supports adding sleep as a bounded use case to Brali's existing conservative PMR protocol. It does not establish an optimal timer, session count or frequency, and the evidence should be described as subjective sleep-quality evidence in the studied clinical populations rather than a universal sleep effect.",
              "limitations": [
                "Between-study heterogeneity was very high (I2 85.5%) and was not explained by the reported subgroup analyses.",
                "The included trials involved diverse clinical populations, limiting direct generalization to healthy adults or any one condition.",
                "Sleep quality was assessed with the subjective PSQI rather than objective sleep measures.",
                "Intervention duration, frequency, session count and protocol details varied substantially across studies.",
                "Egger's test suggested possible publication bias; trim-and-fill reduced the pooled estimate while retaining the direction.",
                "Only 14 studies were available, limiting subgroup and meta-regression precision.",
                "The age >=55 subgroup estimate was imprecise, and the formal between-subgroup difference was not significant.",
                "The review did not establish long-term persistence of benefit after PMR stopped."
              ],
              "source_url": "https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2026.1906525/full"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "cognitive-bias-check",
      "category": "thinking",
      "language": "en",
      "query": "How can I notice cognitive biases before making a decision?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "cognitive-biases"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "assumed-similarity-bias-check",
          "avoid-gamblers-fallacy-trust-the-odds",
          "backfire-effect-coach",
          "10-minute-stress-walk",
          "active-listening-meeting-notes"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "cognitive-biases",
          "decision-making",
          "emotion-regulation"
        ],
        "protocol_slugs": [
          "consider-the-opposite",
          "assumed-similarity-bias-check",
          "active-listening-meeting-notes",
          "anticipate-sudden-trend-shifts",
          "ai-decision-alternatives"
        ],
        "evidence_decision_ids": [
          "debiasing-education-transfer-boundary-2025",
          "consider-opposite-social-judgment-boundary-1984",
          "empathic-paraphrasing-immediate-emotion-boundary-2012"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "consider-the-opposite",
              "canonical_id": "brali:protocol:consider-the-opposite",
              "action": "Write your current conclusion in one sentence. Then ask: what evidence, mechanism, or alternative explanation could make the opposite conclusion reasonable? Generate a concrete alternative rather than telling yourself to be objective. Check whether your decision would change if that alternative were true. Use this on decisions where biased assimilation or one-sided hypothesis testing is a real risk; do not turn it into endless doubt about routine choices.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "slug": "assumed-similarity-bias-check",
              "canonical_id": "brali:protocol:assumed-similarity-bias-check",
              "action": "When interacting with others: - Ask: \"What’s unique about this person’s perspective or experience?\" - Listen actively: Avoid projecting your own traits or beliefs onto them. - Reflect: Notice when you assume similarities without evidence and correct yourself. Example: Don’t assume a coworker has the same work style as you; ask about their preferences and needs.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "active-listening-meeting-notes",
              "canonical_id": "brali:protocol:active-listening-meeting-notes",
              "action": "To stay sharp: - Take notes: Write down key points from the person speaking before you. - Breathe and listen: Avoid rehearsing your own response while someone else is speaking. - Repeat mentally: After someone speaks, quickly repeat their main point in your head. Example: In a team meeting, note what the person before you says and reference it when it’s your turn.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "anticipate-sudden-trend-shifts",
              "canonical_id": "brali:protocol:anticipate-sudden-trend-shifts",
              "action": "When predicting trends: - Ask yourself: \"Could there be sudden changes or breaks in this trend?\" - Prepare for shifts: Always consider outliers and unexpected events in planning. - Look for weak signals: Early signs of change can help you adapt quickly. Example: A stock steadily rising? Plan for a downturn just as much as for continued growth.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-decision-alternatives",
              "canonical_id": "brali:protocol:ai-decision-alternatives",
              "action": "Write your objective, constraints, and current leading option. Ask AI for materially different alternatives, missing criteria, and the strongest case against your favorite. Verify any external facts, then compare options using your own criteria and downside limits. Do not ask the model to hide the trade-off inside a single confident recommendation.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41598-024-60220-5"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "debiasing-education-transfer-boundary-2025",
              "decision": "support-existing",
              "supported_claim": "Debiasing education can produce a small average improvement on targeted bias tasks. This supports modest expectations for a consider-the-opposite check while making the boundary explicit: depth of learning and transfer to meaningful real-world decisions remain uncertain.",
              "limitations": [
                "All included studies were rated unclear or high risk of bias.",
                "There was some evidence of publication bias.",
                "Interventions, bias targets and outcome measures were highly heterogeneous.",
                "Transfer beyond explicitly trained tasks was limited or uncertain in many studies.",
                "The pooled effect represents diverse educational interventions rather than the consider-the-opposite strategy alone."
              ],
              "source_url": "https://www.nature.com/articles/s41562-025-02253-y"
            },
            {
              "id": "consider-opposite-social-judgment-boundary-1984",
              "decision": "propose-protocol",
              "supported_claim": "When a judgment is vulnerable to one-sided evidence processing, deliberately generating an opposed possibility can reduce bias on some tasks more effectively than simply telling oneself to be fair or unbiased. Brali can use this as a concrete pre-decision check while preserving the possibility that the original conclusion remains correct.",
              "limitations": [
                "Classic laboratory/social-judgment evidence from undergraduate samples.",
                "Only two focal task domains were tested in the original article.",
                "Long-term persistence and broad transfer were not established.",
                "The authors note that considering the opposite can in some circumstances overweight disconfirming evidence and create a different form of partiality.",
                "Demand characteristics and task-specific effects remain possible."
              ],
              "source_url": "https://doi.org/10.1037/0022-3514.47.6.1231"
            },
            {
              "id": "empathic-paraphrasing-immediate-emotion-boundary-2012",
              "decision": "support-existing",
              "supported_claim": "In this small nonclinical social-conflict experiment, a trained interviewer used a behavior closely matching Brali's core correction loop: summarize the speaker's facts, feelings and priorities, then ask whether the understanding is accurate. Participants reported less negative immediate emotion after paraphrasing than after silent note-taking. This supports retaining paraphrase plus explicit correction as a bounded practice behavior, with the studied outcome and setting stated precisely.",
              "limitations": [
                "Only twenty participants contributed self-report data, making results vulnerable to individual variation and unsuitable for broad population estimates.",
                "One female interviewer with approximately 190 hours of conflict-resolution training delivered every paraphrase, so listener skill and delivery cannot be separated from the technique.",
                "The control condition was silent note-taking rather than another spoken response; differences may partly reflect receiving a verbal response rather than paraphrasing specifically.",
                "Participants may have interpreted note-taking as judgment despite the study explanation, potentially biasing the comparison.",
                "Only immediate reactions were measured; the authors explicitly described longer-term emotion resolution as speculative.",
                "Conflicts involving physical or psychological violence were excluded, so the result must not be applied as ordinary advice in unsafe or abusive situations.",
                "The study assessed emotional valence and arousal, not whether the listener understood more accurately or whether the conflict outcome improved."
              ],
              "source_url": "https://doi.org/10.3389/fpsyg.2012.00482"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "active-listening-exact",
      "category": "communication",
      "language": "en",
      "query": "Show me active listening exercises",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "listening"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "active-listening-exercises"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "active-listening-exercises",
          "10-minute-language-microsprints",
          "active-listening-meeting-notes",
          "active-recall-test-yourself",
          "ai-conversation-rehearsal"
        ],
        "topic_hit": false,
        "protocol_hit": true,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "listening",
          "memory"
        ],
        "protocol_slugs": [
          "ask-better-interview-questions",
          "active-recall-test-yourself",
          "avoid-list-interference-memory-retention",
          "active-listening-exercises",
          "ai-learning-attempt-feedback-retest"
        ],
        "evidence_decision_ids": [
          "retrieval-procedural-application-2026",
          "testing-versus-restudy-retention-boundary-2014",
          "wakeful-rest-complex-learning-challenge-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ask-better-interview-questions",
              "canonical_id": "brali:protocol:ask-better-interview-questions",
              "action": "Explain why you are asking and get permission. Ask the person to walk through the most recent relevant example. Use one question at a time about sequence, context, tools, decisions and uncertainty. Summarize what you heard and invite correction.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "active-listening-exercises",
              "canonical_id": "brali:protocol:active-listening-exercises",
              "action": "In one conversation, wait until the other person finishes a thought, paraphrase the main point in your own words, invite correction, and ask one open question before offering advice or changing the topic.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "retrieval-procedural-application-2026",
              "decision": "challenge-existing",
              "supported_claim": "Adding retrieval practice to worked examples improved long-term retention of the spelling rules in this classroom study, but the combined group did not outperform worked examples alone on correct rule application after one week.",
              "limitations": [
                "The sample was 105 fourth-grade pupils learning Dutch verb spelling in a small number of intact classes and schools.",
                "The comparison tested retrieval practice added to worked examples, not retrieval practice in isolation.",
                "The target was one form of procedural knowledge and the study does not establish identical boundaries in other domains or age groups.",
                "The result supports separating retention from application outcomes, not concluding that retrieval practice is ineffective for complex learning generally."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/pii/S0959475226000629"
            },
            {
              "id": "testing-versus-restudy-retention-boundary-2014",
              "decision": "support-existing",
              "supported_claim": "Across the included testing-versus-restudy literature, retrieval testing produced better later retention on average than equivalent-duration restudy. The pooled random-effects estimate was positive, but effects varied substantially across studies and testing conditions. This supports Brali's bounded recommendation to retrieve information from memory, check the answer, and use that loop when retention is the target.",
              "limitations": [
                "Between-study heterogeneity in the primary analysis was very high, so the pooled mean does not describe one uniform effect across designs or contexts.",
                "The review deliberately restricted the quantitative synthesis to testing-versus-equivalent-restudy contrasts and does not represent every retrieval-practice paradigm.",
                "Testing benefits varied with methodological factors including initial test format, retention interval and feedback.",
                "Many contributing studies used controlled experimental learning tasks and college samples, which limits direct extrapolation to every real-world learning context.",
                "The review's main outcome is retention; it does not establish that remembered information can be applied correctly in a novel task or procedure.",
                "Theoretical findings did not provide one complete cohesive mechanism for all testing phenomena, so Brali should avoid mechanism decoration.",
                "A positive average effect does not imply that every individual effect was positive; the reviewed distribution included negative and null estimates."
              ],
              "source_url": "https://doi.org/10.1037/a0037559"
            },
            {
              "id": "wakeful-rest-complex-learning-challenge-2026",
              "decision": "challenge-existing",
              "supported_claim": "The existing Brali wakeful-rest protocol must not promise better comprehension or long-term retention for complex educational reading. In this more ecologically valid randomized study, eight minutes of post-reading wakeful rest did not produce a consistent advantage over social media, math or similar-text reading on immediate or one-week factual and conceptual tests. Keep the protocol narrow: it is an optional declarative-memory tactic, not a general learning upgrade.",
              "limitations": [
                "The study used one expository learning task in a university-student population.",
                "The wakeful-rest condition used an autogenic-training implementation and therefore does not represent every possible quiet-rest procedure.",
                "Only one eight-minute post-learning duration was tested.",
                "The delayed test used different but conceptually aligned questions rather than identical items from the immediate test.",
                "A single null study cannot establish absence of a wakeful-rest effect across all complex learning tasks or populations."
              ],
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC12945945/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "clear-message",
      "category": "communication",
      "language": "en",
      "query": "Show me a 5W1H communication checklist for a clear work message",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "clear-communication"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "5w1h-communication-checklist"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "2-minute-desk-tidy-timer",
          "3-3-3-workday-planner",
          "5w1h-communication-checklist",
          "ai-test-case-generator",
          "savor-positive-moment"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "clear-communication",
          "work-systems",
          "planning-prioritization"
        ],
        "protocol_slugs": [
          "5w1h-communication-checklist",
          "build-your-support-network",
          "clarity-ladder-analyzer",
          "leave-an-app-review",
          "ai-customer-reply-coach"
        ],
        "evidence_decision_ids": [
          "supervisory-feedback-characteristics-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "5w1h-communication-checklist",
              "canonical_id": "brali:protocol:5w1h-communication-checklist",
              "action": "Review an important operational message using six questions: Who? What? Where? When? Why? How? Add only the missing details the reader needs to act, then send the shortest version that is still clear.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "build-your-support-network",
              "canonical_id": "brali:protocol:build-your-support-network",
              "action": "Write down one current need, decide what kind of help would actually be useful, identify a person or community that is appropriate for that request, ask specifically and make declining easy, respect the answer, and record whether you need a different resource or a different kind of help next time.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "clarity-ladder-analyzer",
              "canonical_id": "brali:protocol:clarity-ladder-analyzer",
              "action": "Write the main point in plain language, choose an example, observation, step or verified figure that illustrates it, explain the connection in one sentence, and remove details that do not change understanding or action. If you began with a detail, reverse the sequence by stating the broader point it supports.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "leave-an-app-review",
              "canonical_id": "brali:protocol:leave-an-app-review",
              "action": "Describe one concrete thing that worked or failed, explain the effect it had on your task, and add one specific suggestion if you have one.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-customer-reply-coach",
              "canonical_id": "brali:protocol:ai-customer-reply-coach",
              "action": "Provide only information you are allowed to share, plus the customer question and approved policy or knowledge source. Ask for a concise draft. Check that it answers the real issue, does not invent policy or status, and gives the correct next action before sending.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1093/qje/qjae044"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "supervisory-feedback-characteristics-boundary-2026",
              "decision": "watch",
              "supported_claim": "Across the reviewed workplace literature, supervisor credibility and feedback quality were consistently positively associated with several stages of employee feedback processing. These are useful design hypotheses for Brali feedback protocols, but the synthesis does not establish that changing credibility, specificity or constructiveness will causally improve acceptance, behavior, performance or learning.",
              "limitations": [
                "Most included studies were correlational and relied heavily on self-report, limiting causal inference and raising common-method concerns.",
                "Between-study heterogeneity was very high (I²=95%).",
                "Nine studies had potential quality or risk-of-bias limitations under the review's appraisal.",
                "Only peer-reviewed studies were included, so unpublished and grey literature were excluded.",
                "Context variables such as delivery channel, feedback frequency and organizational climate were too inconsistently reported for robust meta-regression.",
                "Some characteristics and processing facets were represented by only a small number of studies, constraining fine-grained conclusions."
              ],
              "source_url": "https://link.springer.com/article/10.1007/s12144-026-09323-y"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "difficult-conversation",
      "category": "communication",
      "language": "en",
      "query": "How can I prepare for a difficult conversation without escalating it?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "difficult-conversations"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "ai-conversation-rehearsal",
          "contextual-language-coach",
          "3-minute-sensory-mindfulness-act",
          "active-listening-exercises",
          "active-listening-meeting-notes"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "difficult-conversations",
          "negotiation",
          "psychological-flexibility"
        ],
        "protocol_slugs": [
          "ai-conversation-rehearsal",
          "ai-customer-reply-coach",
          "create-win-win-outcomes",
          "ancestral-torch-resilience-tracker",
          "contextual-language-coach"
        ],
        "evidence_decision_ids": [
          "chatbot-difficult-conversation-rehearsal-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ai-conversation-rehearsal",
              "canonical_id": "brali:protocol:ai-conversation-rehearsal",
              "action": "State the goal, your opening, and relevant context without unnecessary private detail. Ask AI to play a neutral, skeptical, and confused counterpart. Practise responding with facts, boundaries, and questions. Write one shorter opening after the rehearsal and use that in the real conversation if it still fits.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-customer-reply-coach",
              "canonical_id": "brali:protocol:ai-customer-reply-coach",
              "action": "Provide only information you are allowed to share, plus the customer question and approved policy or knowledge source. Ask for a concise draft. Check that it answers the real issue, does not invent policy or status, and gives the correct next action before sending.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1093/qje/qjae044"
            },
            {
              "slug": "create-win-win-outcomes",
              "canonical_id": "brali:protocol:create-win-win-outcomes",
              "action": "List what each side wants, what each side can trade at relatively low cost, and two or three options that combine those differences. Then test the options with the other side.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ancestral-torch-resilience-tracker",
              "canonical_id": "brali:protocol:ancestral-torch-resilience-tracker",
              "action": "Write one sentence in the form: This is difficult, and the next thing I can control is ___. Keep it specific to the current situation.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "contextual-language-coach",
              "canonical_id": "brali:protocol:contextual-language-coach",
              "action": "Define the scenario and desired outcome, write the core request or response, add likely follow-up phrases and one repair phrase for misunderstanding, rehearse both sides aloud, use the interaction when appropriate, and update the phrase set from the point where you hesitated or needed clarification.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "chatbot-difficult-conversation-rehearsal-2026",
              "decision": "propose-protocol",
              "supported_claim": "For adults already considering an ordinary difficult conversation, a brief chatbot preparation session modestly increased the probability that they reported actually initiating it over the following week: 65% in the preparation condition versus 59% in control. Brali can therefore test a bounded rehearsal protocol whose goal is to reduce the initiation barrier: clarify the goal, draft an opening, role-play a likely response and prepare one or two contingency moves, then return to the real human conversation.",
              "limitations": [
                "The primary real-world outcome was participant self-report rather than independently observed conversation behavior.",
                "The absolute difference in conversation initiation was modest: six percentage points.",
                "The experimental and control chatbot interactions differed in duration, engagement and prompting, so the study does not isolate which preparation component caused the effect.",
                "The sample was limited to US and Canadian Prolific participants, restricting cultural generalizability.",
                "Among participants who did have the difficult conversation, the study found no significant differences in satisfaction, perceived quality, difficulty or thoroughness.",
                "The study followed behavior for roughly one week and does not establish repeated-use or long-term effects.",
                "Preparing the conversation increased negative affect immediately after the exercise, and repeated reliance on a chatbot was not studied.",
                "Safety-sensitive, coercive, legal and high-power-imbalance conversations were not established as appropriate use cases."
              ],
              "source_url": "https://journals.sagepub.com/doi/10.1177/19485506261444068"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "conflict-repair",
      "category": "communication",
      "language": "en",
      "query": "How can I repair a misunderstanding after a conflict?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "conflict-repair"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "relationship-repair-coach",
          "contextual-language-coach",
          "active-recall-test-yourself",
          "break-down-big-problems-triz",
          "create-win-win-outcomes"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "conflict-repair"
        ],
        "protocol_slugs": [
          "relationship-repair-coach",
          "backfire-effect-coach",
          "contextual-language-coach"
        ],
        "evidence_decision_ids": [
          "apology-repair-2021"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "relationship-repair-coach",
              "canonical_id": "brali:protocol:relationship-repair-coach",
              "action": "Wait until you can speak without escalating. Name the specific action or words you are taking responsibility for. Acknowledge the impact as you understand it without using your intention to cancel the other person's experience. Apologize for your part without an 'if'. Offer one concrete repair or behavior change you can actually make, then ask how the other person sees it. Listen to the answer. Do not require forgiveness, agreement, or immediate trust in return.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34162912/"
            },
            {
              "slug": "backfire-effect-coach",
              "canonical_id": "brali:protocol:backfire-effect-coach",
              "action": "Notice that you are defending rather than examining. Pause before replying. Restate the other person's claim in words they accept, then state your own position and uncertainty. Ask what information would change either view. Choose one concrete check, or stop when the exchange is no longer productive or safe.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "contextual-language-coach",
              "canonical_id": "brali:protocol:contextual-language-coach",
              "action": "Define the scenario and desired outcome, write the core request or response, add likely follow-up phrases and one repair phrase for misunderstanding, rehearse both sides aloud, use the interaction when appropriate, and update the phrase set from the point where you hesitated or needed clarification.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "apology-repair-2021",
              "decision": "propose-protocol",
              "supported_claim": "In the controlled experiments, apologies promoted forgiveness and operated through a causal pathway shared with perceived relationship value. This supports apology as one potentially useful repair signal after a workable interpersonal transgression.",
              "limitations": [
                "Controlled experimental transgression paradigms do not capture every feature of real-world repeated or high-stakes conflict.",
                "Forgiveness is only one outcome and is not equivalent to trust, reconciliation, safety, or durable behavior change.",
                "The effect of apology depends on context, relationship value and the nature of the transgression.",
                "The study does not validate the specific wording or step sequence used in Brali's public repair protocol."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34162912/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "movement-break-boundary",
      "category": "movement",
      "language": "en",
      "query": "What is the scientifically proven best break interval for cognition while sitting?",
      "mode": "bounded-evidence",
      "expected": {
        "topic_ids": [
          "movement"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": [
          "movement-breaks-cognition-2026"
        ]
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "25-minute-pomodoro-focus-sprints",
          "4-minute-hiit-tabata-workout",
          "spaced-recall-coach",
          "walking-meeting-assistant",
          "10-minute-stress-walk"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": true,
        "topic_ids": [
          "movement",
          "recovery-energy"
        ],
        "protocol_slugs": [],
        "evidence_decision_ids": [
          "movement-breaks-cognition-2026",
          "sitting-breaks-cognition-boundary-2026",
          "distributed-practice-verbal-recall-boundary-2006"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": 1,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 0.8571,
        "answer_packet": {
          "protocols": [],
          "evidence_boundaries": [
            {
              "id": "movement-breaks-cognition-2026",
              "decision": "watch",
              "supported_claim": "Across the included trials, interruptions to prolonged sitting were associated with small acute improvements in executive function and memory on average. Evidence for global cognition, information processing, and attention was insufficient or inconsistent, and certainty was low for most outcomes.",
              "limitations": [
                "Only 21 randomized crossover trials with 433 participants were included.",
                "GRADE certainty was low for executive function, memory, information processing, and attention and very low for global cognition.",
                "No clear improvement was found for global cognition, information processing, or attention.",
                "Moderator and protocol-parameter findings were exploratory and the authors explicitly state they are insufficient for definitive prescriptive recommendations.",
                "The outcomes were acute cognitive effects, so the source does not establish long-term cognitive benefit from a break routine."
              ],
              "source_url": "https://link.springer.com/article/10.1186/s12966-026-01953-6"
            },
            {
              "id": "sitting-breaks-cognition-boundary-2026",
              "decision": "watch",
              "supported_claim": "The existing Brali sitting-break protocol has limited additional cognitive support: across randomized crossover studies, interrupting prolonged sitting was associated with small acute improvements in executive function and memory. Because certainty was low and attention/information-processing results were not clear, cognition should remain a secondary possible benefit rather than the reason or promise for the protocol.",
              "limitations": [
                "Only 21 crossover trials with 433 total participants were included, leaving many domain and subgroup estimates based on small evidence bases.",
                "GRADE certainty was low for executive function, memory, information processing and attention and very low for global cognition.",
                "The evidence concerns acute experimental responses, not long-term cognitive outcomes, productivity or real-world work performance.",
                "Intervention protocols varied substantially in frequency, intensity, duration and mode.",
                "Subgroup and meta-regression findings were exploratory and should not be converted into individualized dosing rules.",
                "Some apparently favorable domain estimates may be sensitive to the small number of studies and heterogeneous cognitive tasks."
              ],
              "source_url": "https://link.springer.com/article/10.1186/s12966-026-01953-6"
            },
            {
              "id": "distributed-practice-verbal-recall-boundary-2006",
              "decision": "support-existing",
              "supported_claim": "For verbal material measured by later recall, separating repeated study episodes by a meaningful interval generally supports better retention than concentrating the same material into massed study. The spacing associated with the best later recall tended to increase as the intended retention interval increased. For material that must be retained over months or years, the reviewed evidence supports distributing study across days or longer rather than completing all review in one sitting or one day. These findings justify the rewritten Brali action only within a verbal-recall boundary and without one universal schedule.",
              "limitations": [
                "The synthesis was restricted to verbal memory tasks measured by recall and deliberately excluded recognition, frequency judgments and the heterogeneous skill-learning literature.",
                "The evidence base was dominated by young adults; the authors reported very little middle-aged and older-adult evidence and insufficient long-term child data for confident generalization.",
                "Many studies did not report the variance data needed for effect-size calculation, so several analyses relied on accuracy differences and included fewer effect-size estimates.",
                "Published null findings may be underrepresented because of the file-drawer problem.",
                "Study materials, presentation schedules, retention intervals and experimental procedures varied substantially.",
                "Some historical studies confounded longer spacing with more relearning trials; the review examined this problem but could not remove every design limitation from the literature.",
                "Binning inter-study and retention intervals supported broad patterns but reduced the ability to recommend exact intervals.",
                "The useful interval depends jointly on spacing and the later retention target, and the authors stated that exact long-term optimization could not be specified with certainty.",
                "Evidence comparing expanding and fixed schedules was sparse and inconsistent, with large between-study variability.",
                "The synthesis addresses later recall, not broader comprehension, transfer, motivation, study adherence or real-world performance.",
                "The review was published in 2006 and should be rechecked against newer syntheses before adding more precise scheduling claims."
              ],
              "source_url": "https://doi.org/10.1037/0033-2909.132.3.354"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "idea-generation",
      "category": "creativity",
      "language": "en",
      "query": "How can a short timebox help me produce a rough first pass before judging ideas?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "idea-generation"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "30-minute-deadline-sprints"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "30-minute-deadline-sprints",
          "ai-rubric-critique",
          "ai-idea-divergence-pass",
          "ask-a-peer-to-proofread",
          "10-minute-language-microsprints"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "idea-generation"
        ],
        "protocol_slugs": [
          "30-minute-deadline-sprints",
          "ai-idea-divergence-pass",
          "bold-brainstorm-kickoff",
          "brainwriting-group-idea-generation",
          "ai-rubric-critique"
        ],
        "evidence_decision_ids": [
          "human-first-ai-collaboration-boundary-2026",
          "procrastination-treatment-exact-protocol-boundary-2018",
          "vocabulary-pretesting-guess-feedback-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "30-minute-deadline-sprints",
              "canonical_id": "brali:protocol:30-minute-deadline-sprints",
              "action": "Pick a low-risk task where an imperfect first pass is useful. Set a short timebox, define the minimum useful output, and decide what you will leave for later. Stop or reassess when the timebox ends.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-idea-divergence-pass",
              "canonical_id": "brali:protocol:ai-idea-divergence-pass",
              "action": "State the problem and constraints. Ask for several directions that must differ from one another in a named way. Cluster similar answers, discard obvious filler, and evaluate the remaining options against real constraints and evidence.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41562-024-01953-1"
            },
            {
              "slug": "bold-brainstorm-kickoff",
              "canonical_id": "brali:protocol:bold-brainstorm-kickoff",
              "action": "State the real problem. Choose an extreme seed that changes one constraint, customer, resource or rule. During a short generation round, list consequences and variants without arguing with the seed. Remove the seed, restore actual constraints and choose one feasible fragment to test or develop.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "brainwriting-group-idea-generation",
              "canonical_id": "brali:protocol:brainwriting-group-idea-generation",
              "action": "Give the group one clear question, let everyone write ideas silently, pass the ideas to another person to add or change them, and delay discussion until the group has built several options.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-rubric-critique",
              "canonical_id": "brali:protocol:ai-rubric-critique",
              "action": "Write three to six criteria that describe a good result. Give the model the draft and rubric, ask for evidence from the draft for each criticism, and request no rewrite on the first pass. Triage the feedback, then make the changes you agree with.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "human-first-ai-collaboration-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "When preserving a worker's sense of authorship, competence and connection to the task matters, a meaningful human first pass before AI refinement is a reasonable workflow to test. In the randomized writing task, direct copy-and-paste use produced lower psychological ownership and meaningfulness than both no-AI and human-first collaboration, while the human-first condition was comparable to no-AI on those outcomes.",
              "limitations": [
                "The causal experiment used short occupation-specific writing tasks rather than the full range of knowledge work.",
                "The passive condition was intentionally extreme: participants used AI-generated content directly without modification.",
                "The active condition tested one workflow only: human draft first, then AI refinement.",
                "The study did not instrument the external AI tool deeply enough to analyze prompt content and detailed interaction sequences.",
                "Baseline AI skill and confidence were not comprehensively modeled.",
                "The broader follow-up survey was correlational and cannot establish direction of causality."
              ],
              "source_url": "https://doi.org/10.1038/s41598-026-42312-6"
            },
            {
              "id": "procrastination-treatment-exact-protocol-boundary-2018",
              "decision": "challenge-existing",
              "supported_claim": "Across the reviewed randomized comparisons, psychological treatments targeting procrastination had a small average post-treatment benefit on self-reported procrastination relative to inactive controls, but effects varied substantially between studies and the evidence base had important quality limitations. This supports treating task initiation as a legitimate intervention target, not claiming that one brief Brali routine or one component has been validated.",
              "limitations": [
                "The pooled overall effect was small and showed significant between-study heterogeneity, so the average should not be treated as one stable effect across interventions or populations.",
                "The review identified some risk of bias in every included study; blinding-related domains were frequently rated high risk.",
                "Primary procrastination measures varied, and several assessed academic rather than general procrastination.",
                "The included studies used different interventions, delivery formats, durations and participant populations, preventing component-level attribution.",
                "The review analyzed post-treatment outcomes because too few studies reported follow-up data, limiting conclusions about durability.",
                "Secondary outcomes were too heterogeneous to aggregate.",
                "The stronger cognitive-behavioral subgroup estimate depended on excluding one small outlying study and represented only three of the four cognitive-behavioral studies.",
                "The source did not test the exact five-minute interval, the exact Brali action sequence or the protocol's observable-review checkpoint."
              ],
              "source_url": "https://doi.org/10.3389/fpsyg.2018.01588"
            },
            {
              "id": "vocabulary-pretesting-guess-feedback-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For adults learning new, concrete second-language vocabulary paired with images, adding a forced multiple-choice guess before immediately revealing the correct pairing can produce a modest short-delay memory advantage over simply reading the pair. A bounded Brali protocol may therefore use a cue → guess → immediate correct feedback → later recall sequence for vocabulary practice, while keeping the claim limited to this learning format and time horizon.",
              "limitations": [
                "All criterial tests followed a short same-session distractor period rather than a delayed retention interval, so long-term retention was not established.",
                "The studies used concrete Spanish nouns paired with images; vocabulary type and language-learning context were narrow.",
                "Participants were adult Prolific users from English-speaking countries and exclusions were substantial in several experiments.",
                "Pretesting trials included up to 8 seconds of cue-only guessing before the same 5 seconds of correct pair exposure used in reading, so total cue exposure was longer in the pretesting condition.",
                "Initial guess accuracy was about 35–38%, raising the possibility of some prior familiarity despite exclusions; the authors note that benefits also appeared for incorrectly guessed items.",
                "Learning condition was primarily within-subjects, and the authors identify between-subject replication as a future research need.",
                "Multiple-choice benefits were not significant in Experiment 3, so recognition effects were less uniform than cued-recall effects."
              ],
              "source_url": "https://link.springer.com/article/10.1186/s41235-026-00708-y"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "brainwriting-exact",
      "category": "creativity",
      "language": "en",
      "query": "Show me a brainwriting group idea generation protocol",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "idea-generation"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "brainwriting-group-idea-generation"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "batch-non-urgent-notifications",
          "bold-brainstorm-kickoff",
          "brainwriting-group-idea-generation",
          "4-minute-tabata-hiit-timer",
          "active-listening-meeting-notes"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "idea-generation",
          "writing"
        ],
        "protocol_slugs": [
          "bold-brainstorm-kickoff",
          "brainwriting-group-idea-generation",
          "ai-idea-divergence-pass",
          "biomimicry-creative-problem-solving",
          "creativity-within-constraints"
        ],
        "evidence_decision_ids": [
          "walking-divergent-thinking-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "bold-brainstorm-kickoff",
              "canonical_id": "brali:protocol:bold-brainstorm-kickoff",
              "action": "State the real problem. Choose an extreme seed that changes one constraint, customer, resource or rule. During a short generation round, list consequences and variants without arguing with the seed. Remove the seed, restore actual constraints and choose one feasible fragment to test or develop.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "brainwriting-group-idea-generation",
              "canonical_id": "brali:protocol:brainwriting-group-idea-generation",
              "action": "Give the group one clear question, let everyone write ideas silently, pass the ideas to another person to add or change them, and delay discussion until the group has built several options.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-idea-divergence-pass",
              "canonical_id": "brali:protocol:ai-idea-divergence-pass",
              "action": "State the problem and constraints. Ask for several directions that must differ from one another in a named way. Cluster similar answers, discard obvious filler, and evaluate the remaining options against real constraints and evidence.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41562-024-01953-1"
            },
            {
              "slug": "biomimicry-creative-problem-solving",
              "canonical_id": "brali:protocol:biomimicry-creative-problem-solving",
              "action": "State what the solution must do without naming the current solution. Choose a biological example from a reliable description. Separate the observed structure or process from the metaphor. Write one transferable principle, generate several applications and test the smallest reversible option within safety and operational constraints.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "creativity-within-constraints",
              "canonical_id": "brali:protocol:creativity-within-constraints",
              "action": "Name the artifact and the part that feels too open, choose one constraint that does not violate the brief, create a reversible variant without adding new constraints midstream, compare the result with the original requirements, and keep only the useful difference.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "walking-divergent-thinking-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "Walking is a defensible context change for the idea-generation phase of creative work. The meta-analysis found moderate-certainty evidence of a large positive effect on divergent thinking, including in randomized-only sensitivity analysis. The same evidence does not establish a benefit for convergent evaluation or selection.",
              "limitations": [
                "Substantial heterogeneity remained unexplained.",
                "Most participants were post-secondary students.",
                "Divergent-thinking measures dominated the literature.",
                "Evidence for convergent thinking was rated very low certainty and was based on few studies.",
                "Several positive studies originated from a limited set of research groups and settings.",
                "The review does not identify an optimal duration or intensity."
              ],
              "source_url": "https://doi.org/10.1371/journal.pone.0347878"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "writing-outline",
      "category": "creativity",
      "language": "en",
      "query": "How can I structure and draft a piece of writing?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "writing"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "structure-writing-with-visuals",
          "ai-professional-writing-draft",
          "context-planner-for-writers",
          "active-recall-test-yourself",
          "ai-customer-reply-coach"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "writing",
          "decision-making"
        ],
        "protocol_slugs": [
          "structure-writing-with-visuals",
          "ai-professional-writing-draft",
          "ask-a-peer-to-proofread",
          "ai-rubric-critique",
          "context-planner-for-writers"
        ],
        "evidence_decision_ids": [
          "structured-peer-feedback-provision-2025",
          "writing-feedback-level-match-2024"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 2,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "structure-writing-with-visuals",
              "canonical_id": "brali:protocol:structure-writing-with-visuals",
              "action": "Name the reader's main task, read the current headings without the body, reorganize sections around that path, replace repeated comparisons with a table when rows share the same attributes, use a diagram for relationships or sequence, and keep images only when they show information that prose would make harder to inspect.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-professional-writing-draft",
              "canonical_id": "brali:protocol:ai-professional-writing-draft",
              "action": "Provide the audience, goal, source facts, must-include points, and constraints. Ask for one draft. Compare it with the source material, remove invented details, restore your own judgment and voice, and send only the version you can defend without the model in the room.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1126/science.adh2586"
            },
            {
              "slug": "ask-a-peer-to-proofread",
              "canonical_id": "brali:protocol:ask-a-peer-to-proofread",
              "action": "Before asking for feedback on a student or learning-focused draft, choose the current revision target. Use a surface pass for grammar, spelling, wording, and similar features; use a deep pass for ideas, organization, coherence, or other meaning-level features. Tell the reviewer which level you want them to focus on, revise against that target, and run a separate pass if the other level also needs work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.learninstruc.2024.101961"
            },
            {
              "slug": "ai-rubric-critique",
              "canonical_id": "brali:protocol:ai-rubric-critique",
              "action": "Write three to six criteria that describe a good result. Give the model the draft and rubric, ask for evidence from the draft for each criticism, and request no rewrite on the first pass. Triage the feedback, then make the changes you agree with.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "context-planner-for-writers",
              "canonical_id": "brali:protocol:context-planner-for-writers",
              "action": "Write the primary reader, publication surface, desired reader action or understanding, required tone or format constraints, facts that must be checked, and any detail that should be omitted for privacy, confidentiality or scope. Then build the outline from that brief.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "structured-peer-feedback-provision-2025",
              "decision": "support-existing",
              "supported_claim": "In the reviewed educational literature, adding instructional support to peer feedback improved the peer-feedback process overall, and support aimed at the person providing feedback was associated with better feedback-provision quality. This supports Brali's existing educational writing protocol in making the review target explicit and giving the peer reviewer a small amount of structure instead of requesting vague general feedback.",
              "limitations": [
                "Only 32 journal studies met the inclusion criteria, limiting fine-grained moderator analysis.",
                "Only three studies with 14 effect sizes examined feedback-reception support, making conclusions about that phase underpowered and unstable.",
                "The authors had to collapse specific supports such as rubrics, sentence starters and guiding questions into broader categories because each was represented by too few studies.",
                "Study heterogeneity was high, and publication-bias tests were mixed: Egger's test was significant whereas Begg's test was not and the funnel plot was relatively symmetrical.",
                "The meta-analysis concerns educational peer-feedback processes and should not be treated as direct evidence for all professional writing or workplace review contexts.",
                "Coarse outcome categories do not reveal which exact support mechanism is best for a specific writing task."
              ],
              "source_url": "https://link.springer.com/article/10.1007/s10648-025-10017-3"
            },
            {
              "id": "writing-feedback-level-match-2024",
              "decision": "propose-protocol",
              "supported_claim": "In the reviewed secondary-school and university writing literature, feedback effects differed by the level of writing outcome: surface-level feedback improved surface-level outcomes, deep-level feedback improved deep-level outcomes, and combined surface-and-deep feedback could improve both. This supports matching a feedback request to the revision outcome rather than treating all feedback as interchangeable.",
              "limitations": [
                "The evidence concerns educational writing by secondary-school and university or college learners rather than all writing contexts.",
                "Effects varied across L1, L2, and foreign-language learners, so a single population-wide estimate can conceal important differences.",
                "The included papers used experimental or quasi-experimental designs but varied in feedback treatment, source, duration, outcome measurement, and instructional context.",
                "Only about one-third of the treatment/control comparisons examined maintenance effects, limiting conclusions about durability.",
                "Many studies did not report enough learner-level variables, and evidence was comparatively scarce for some learner groups and treatment-outcome combinations.",
                "Meta-analytic categories for surface, deep, combined, and feedback source are useful abstractions but do not establish one universally optimal revision workflow."
              ],
              "source_url": "https://doi.org/10.1016/j.learninstruc.2024.101961"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "work-system",
      "category": "work",
      "language": "en",
      "query": "How can I make a recurring work process more reliable?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "work-systems"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "biomimicry-creative-problem-solving",
          "2-minute-desk-tidy-timer",
          "25-minute-pomodoro-focus-sprints",
          "3-3-3-workday-planner",
          "30-min-eye-break-tracker"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "work-systems",
          "planning-prioritization",
          "revision-feedback"
        ],
        "protocol_slugs": [
          "ai-task-frontier-test",
          "deliberate-email-checking-windows",
          "3-3-3-workday-planner",
          "attention-to-detail-trainer",
          "brag-doc-wins-hub"
        ],
        "evidence_decision_ids": [
          "email-checking-frequency-stress-boundary-2015"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ai-task-frontier-test",
              "canonical_id": "brali:protocol:ai-task-frontier-test",
              "action": "Choose a small sample containing easy, ordinary, and awkward cases. Define what counts as acceptable before running the model. Compare AI output with the source or a trusted human result, record failure patterns, and decide which cases can be assisted, which need review, and which should stay human-first.",
              "evidence_state": "reviewed",
              "source_url": "https://aiinstitute.hbs.edu/navigating-the-jagged-technological-frontier/"
            },
            {
              "slug": "deliberate-email-checking-windows",
              "canonical_id": "brali:protocol:deliberate-email-checking-windows",
              "action": "Choose a small number of email windows that still meet the real response expectations of your work and personal life. Close or hide the inbox between those windows and disable nonessential email alerts. Keep a separate urgent channel for issues that genuinely cannot wait. Do not treat three checks per day as a universal rule; that was the experimental condition, not an optimized schedule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            },
            {
              "slug": "3-3-3-workday-planner",
              "canonical_id": "brali:protocol:3-3-3-workday-planner",
              "action": "Divide the workday into up to three broad blocks. Give each block one to three clear outcomes that fit the time you actually have. Replan when meetings, urgent work, or delays change the day.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "attention-to-detail-trainer",
              "canonical_id": "brali:protocol:attention-to-detail-trainer",
              "action": "Choose an artifact or environment where errors matter, write a short checklist from known failure modes, inspect once for completeness and once from a different perspective such as data consistency or user flow, record anomalies, fix them, and update the checklist when a new recurring miss appears.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "brag-doc-wins-hub",
              "canonical_id": "brali:protocol:brag-doc-wins-hub",
              "action": "After a meaningful piece of work, add a short entry with the situation, your contribution, the concrete result or current state, and one or two tags. During a review, retrieve entries that match the purpose and verify any figures or claims before reusing them.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "email-checking-frequency-stress-boundary-2015",
              "decision": "propose-protocol",
              "supported_claim": "When role expectations allow it, checking email less frequently than one's normal pattern can reduce daily stress. A practical Brali implementation is to use deliberate email windows while keeping a separate route for genuinely urgent work. The study supports less-frequent checking, not three checks per day as an optimized rule.",
              "limitations": [
                "Checking frequency was self-reported rather than objectively logged.",
                "The study explored a broad set of outcomes, increasing the chance of isolated significant results.",
                "Direct causal evidence was strongest for stress; broader wellbeing links were indirect correlational analyses through stress.",
                "There was no passive measurement-only control condition.",
                "The experiment did not control the response expectations imposed by participants' workplaces and contacts."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "personal-finance",
      "category": "work",
      "language": "en",
      "query": "How can I improve basic personal finance planning and money habits?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "personal-finance"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "3-minute-sensory-mindfulness-act",
          "ai-rubric-critique",
          "ai-structured-extraction-check",
          "anticipate-sudden-trend-shifts",
          "attention-to-detail-trainer"
        ],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0.6667
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "personal-finance",
          "habits-consistency",
          "health-basics"
        ],
        "protocol_slugs": [
          "automatic-paycheck-to-savings",
          "temptation-bundling",
          "if-then-rules-productivity",
          "visible-progress-monitoring",
          "coot-ai-growth-journal"
        ],
        "evidence_decision_ids": [
          "temptation-bundling-exercise-boundary-2020"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "automatic-paycheck-to-savings",
              "canonical_id": "brali:protocol:automatic-paycheck-to-savings",
              "action": "Review upcoming essential expenses, choose a small transfer amount you can change or cancel, schedule it after income arrives, and check that it does not create a shortfall.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "slug": "if-then-rules-productivity",
              "canonical_id": "brali:protocol:if-then-rules-productivity",
              "action": "Choose one repeated situation, describe the cue in observable terms, choose a small action you can safely perform when the cue appears, add a fallback for a common exception, use the rule when the situation occurs, and revise or retire it when it stops fitting the work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1080/10463283.2024.2334563"
            },
            {
              "slug": "visible-progress-monitoring",
              "canonical_id": "brali:protocol:visible-progress-monitoring",
              "action": "Choose one goal where feedback can still change what you do. Pick one observable indicator that is close to the real outcome or behavior you care about. Record it at checkpoints that match the task, compare it with the target or previous checkpoint, and decide one adjustment. Prefer a small trace you will actually use over a dashboard full of decorative metrics.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/bul0000025"
            },
            {
              "slug": "coot-ai-growth-journal",
              "canonical_id": "brali:protocol:coot-ai-growth-journal",
              "action": "State one concrete goal or decision, provide the relevant constraints and what you already know, ask the AI for a small set of distinct options with assumptions and failure modes, challenge anything vague, verify factual claims that matter, choose one reversible next action yourself, and record what would cause you to change course.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": [
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "strategy-choice",
      "category": "work",
      "language": "en",
      "query": "How can I decide what not to do when choosing a strategy?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "strategy"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "25-minute-pomodoro-focus-sprints",
          "ai-code-scaffold-test-loop",
          "ai-idea-divergence-pass",
          "ai-rubric-critique",
          "ai-task-frontier-test"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "strategy",
          "cardiovascular-health",
          "metacognition"
        ],
        "protocol_slugs": [
          "ai-task-frontier-test",
          "create-win-win-outcomes",
          "cardio-health-daily-habits",
          "prequestions-before-learning",
          "refocus-present-stop-mind-wandering"
        ],
        "evidence_decision_ids": [
          "prequestions-targeted-learning-boundary-2025"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 1,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ai-task-frontier-test",
              "canonical_id": "brali:protocol:ai-task-frontier-test",
              "action": "Choose a small sample containing easy, ordinary, and awkward cases. Define what counts as acceptable before running the model. Compare AI output with the source or a trusted human result, record failure patterns, and decide which cases can be assisted, which need review, and which should stay human-first.",
              "evidence_state": "reviewed",
              "source_url": "https://aiinstitute.hbs.edu/navigating-the-jagged-technological-frontier/"
            },
            {
              "slug": "create-win-win-outcomes",
              "canonical_id": "brali:protocol:create-win-win-outcomes",
              "action": "List what each side wants, what each side can trade at relatively low cost, and two or three options that combine those differences. Then test the options with the other side.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "cardio-health-daily-habits",
              "canonical_id": "brali:protocol:cardio-health-daily-habits",
              "action": "Look at the coming week, mark the activity you already do, then add realistic walking, moderate or vigorous activity, and strength sessions where they fit. Build gradually rather than treating the guideline as a one-day target.",
              "evidence_state": "reviewed",
              "source_url": "https://www.heart.org/en/healthy-living/healthy-lifestyle/lifes-essential-8/how-to-be-more-active-fact-sheet"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "prequestions-targeted-learning-boundary-2025",
              "decision": "propose-protocol",
              "supported_claim": "Prequestions can improve later learning of the specific information they target. The meta-analysis reported g = 0.66 for prequestioned information and g = 0.01 for non-prequestioned information; feedback alongside prequestions was associated with stronger targeted learning.",
              "limitations": [
                "The average benefit is specific to prequestioned information rather than a broad benefit to all content.",
                "The meta-analysis does not establish one universal question count, timing rule, or content format.",
                "Effect sizes summarize heterogeneous studies and do not guarantee the same result for an individual learning session."
              ],
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "digital-attention",
      "category": "digital",
      "language": "en",
      "query": "Notifications and feeds keep distracting me. How can I reduce digital distraction?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "digital-attention"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "batch-non-urgent-notifications",
          "30-minute-deadline-sprints",
          "5-minute-meditation-habit-tracker",
          "ab-test-learning-loop",
          "abdominal-breathing-stress-relief"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "digital-attention",
          "digital-wellbeing",
          "task-initiation"
        ],
        "protocol_slugs": [
          "batch-non-urgent-notifications",
          "deliberate-email-checking-windows",
          "grayscale-phone-friction",
          "30-min-eye-break-tracker",
          "20-20-20-eye-break-reminder"
        ],
        "evidence_decision_ids": [
          "notification-blocking-workday-boundary-2023",
          "notification-batching-boundary-2019",
          "email-checking-frequency-stress-boundary-2015"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "deliberate-email-checking-windows",
              "canonical_id": "brali:protocol:deliberate-email-checking-windows",
              "action": "Choose a small number of email windows that still meet the real response expectations of your work and personal life. Close or hide the inbox between those windows and disable nonessential email alerts. Keep a separate urgent channel for issues that genuinely cannot wait. Do not treat three checks per day as a universal rule; that was the experimental condition, not an optimized schedule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            },
            {
              "slug": "grayscale-phone-friction",
              "canonical_id": "brali:protocol:grayscale-phone-friction",
              "action": "Enable grayscale on your phone and leave the rest of your normal setup alone for a bounded trial. Use the phone when you actually need it; the point is not abstinence. Notice whether automatic checking becomes less attractive. If color is important for navigation, accessibility, work, photos, or another real task, switch color back on for that task instead of treating grayscale as a purity test.",
              "evidence_state": "reviewed",
              "source_url": "https://www.frontiersin.org/journals/digital-health/articles/10.3389/fdgth.2026.1816095/full"
            },
            {
              "slug": "30-min-eye-break-tracker",
              "canonical_id": "brali:protocol:30-min-eye-break-tracker",
              "action": "At a natural breakpoint in a long screen session, look away from the display, relax your posture, and name the next task before returning.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "20-20-20-eye-break-reminder",
              "canonical_id": "brali:protocol:20-20-20-eye-break-reminder",
              "action": "During long periods of screen use, take a short visual break about every 20 minutes and look at something about 20 feet away for 20 seconds. Use the reminder as a simple screen-rest cue, not as a treatment or a substitute for eye care.",
              "evidence_state": "reviewed",
              "source_url": "https://www.nei.nih.gov/eye-health-information/healthy-vision/how-eyes-work/keep-your-eyes-healthy"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "notification-blocking-workday-boundary-2023",
              "decision": "support-existing",
              "supported_claim": "Disabling automatic notifications can reduce notification interruptions; in this field experiment, fewer interruptions mediated better perceived performance and lower irritation.",
              "limitations": [
                "The study was a one-day field experiment, so it does not establish durable effects of notification blocking over longer work periods or adaptation over time.",
                "Performance and irritation outcomes were self-reported, and the direct intervention effect on performance was not significant; the reported performance pathway was indirect through fewer interruptions.",
                "Work roles differ in response-time obligations, telepressure, fear of missing out and safety or escalation requirements, so an always-off notification rule is not supported.",
                "The study tests communication-application notification interruptions, not every form of digital distraction or every focus-window design."
              ],
              "source_url": "https://doi.org/10.1002/1348-9585.12408"
            },
            {
              "id": "notification-batching-boundary-2019",
              "decision": "propose-protocol",
              "supported_claim": "Predictable batching of non-urgent notifications is a defensible attention-environment experiment. In this field trial, three daily batches reduced perceived interruption and stress relative to usual delivery, while hourly batching changed little and complete notification removal increased anxiety/FoMO. Brali should preserve urgent exceptions and treat schedule frequency as a user-fit parameter rather than a fixed dose.",
              "limitations": [
                "Single short field experiment rather than a replicated long-term evidence base.",
                "Participants were recruited through an online labor market and may not represent all smartphone users or work contexts.",
                "The intervention used a custom notification-management implementation.",
                "Many psychological outcomes were self-reported and the study intentionally used broad exploratory measurement.",
                "The study compared a small set of schedules and cannot identify an optimal individualized batching frequency."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "id": "email-checking-frequency-stress-boundary-2015",
              "decision": "propose-protocol",
              "supported_claim": "When role expectations allow it, checking email less frequently than one's normal pattern can reduce daily stress. A practical Brali implementation is to use deliberate email windows while keeping a separate route for genuinely urgent work. The study supports less-frequent checking, not three checks per day as an optimized rule.",
              "limitations": [
                "Checking frequency was self-reported rather than objectively logged.",
                "The study explored a broad set of outcomes, increasing the chance of isolated significant results.",
                "Direct causal evidence was strongest for stress; broader wellbeing links were indirect correlational analyses through stress.",
                "There was no passive measurement-only control condition.",
                "The experiment did not control the response expectations imposed by participants' workplaces and contacts."
              ],
              "source_url": "https://doi.org/10.1016/j.chb.2014.11.005"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "information-management",
      "category": "digital",
      "language": "en",
      "query": "How can I capture and retrieve useful information without overbuilding my system?",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "information-management"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "chunk-complex-ideas-for-clarity",
          "active-recall-test-yourself",
          "ai-test-case-generator",
          "batch-non-urgent-notifications",
          "brag-doc-wins-hub"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "information-management",
          "work-systems",
          "attention-focus"
        ],
        "protocol_slugs": [
          "brag-doc-wins-hub",
          "chunk-complex-ideas-for-clarity",
          "ai-source-summary-verification",
          "ai-structured-extraction-check",
          "ai-premortem-risk-planner"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "brag-doc-wins-hub",
              "canonical_id": "brali:protocol:brag-doc-wins-hub",
              "action": "After a meaningful piece of work, add a short entry with the situation, your contribution, the concrete result or current state, and one or two tags. During a review, retrieve entries that match the purpose and verify any figures or claims before reusing them.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "chunk-complex-ideas-for-clarity",
              "canonical_id": "brali:protocol:chunk-complex-ideas-for-clarity",
              "action": "Name the primary audience and the decision or action they need to understand. Group related material by purpose, give each group a descriptive label, order the groups into a main path, and move supporting definitions, examples, calculations or source detail into a layer that remains easy to retrieve.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-source-summary-verification",
              "canonical_id": "brali:protocol:ai-source-summary-verification",
              "action": "Provide only material you are permitted to share. Ask for a summary with sections, explicit uncertainties, and source locations when available. Mark the claims that affect your decision, then open the source and verify those claims directly before reusing them.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1038/s41598-024-60220-5"
            },
            {
              "slug": "ai-structured-extraction-check",
              "canonical_id": "brali:protocol:ai-structured-extraction-check",
              "action": "Define the columns and allowed values before extraction. Ask the model to return null rather than guess, preserve a source snippet or locator for each row, and flag ambiguous cases. Check a random sample plus all high-impact rows against the original text before using the result downstream.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ai-premortem-risk-planner",
              "canonical_id": "brali:protocol:ai-premortem-risk-planner",
              "action": "Describe the plan without sensitive information, ask an AI for plausible ways it could fail, group the suggestions, verify the important ones yourself, and choose mitigations only for risks that survive the check.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "workspace-friction",
      "category": "digital",
      "language": "en",
      "query": "Show me a 2 minute desk tidy timer to reduce workspace friction before I start work",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "workspace"
        ],
        "acceptable_topic_ids": [
          "environment-design"
        ],
        "protocol_slugs": [
          "2-minute-desk-tidy-timer"
        ],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "2-minute-desk-tidy-timer",
          "25-minute-pomodoro-focus-sprints",
          "10-minute-morning-stretch-routine",
          "30-minute-deadline-sprints",
          "4-minute-hiit-tabata-workout"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "workspace",
          "task-initiation",
          "work-systems"
        ],
        "protocol_slugs": [
          "2-minute-desk-tidy-timer",
          "ergonomic-workspace-assessment",
          "if-then-rules-productivity",
          "ready-to-resume-plan",
          "temptation-bundling"
        ],
        "evidence_decision_ids": [
          "ready-to-resume-interruption-boundary-2018",
          "temptation-bundling-exercise-boundary-2020",
          "activity-breaks-postprandial-metabolism-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "2-minute-desk-tidy-timer",
              "canonical_id": "brali:protocol:2-minute-desk-tidy-timer",
              "action": "Set a short timer and tidy one small part of your workspace. Make only obvious moves, then stop before tidying turns into a larger task.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "ergonomic-workspace-assessment",
              "canonical_id": "brali:protocol:ergonomic-workspace-assessment",
              "action": "Choose one frequent task, walk through it slowly, note every unnecessary reach, obstruction, visibility issue, cable move, device swap, or reset, rank the friction points, change one reversible layout variable, use the task again, and keep or undo the change based on what became easier or harder.",
              "evidence_state": "practical",
              "source_url": null
            },
            {
              "slug": "if-then-rules-productivity",
              "canonical_id": "brali:protocol:if-then-rules-productivity",
              "action": "Choose one repeated situation, describe the cue in observable terms, choose a small action you can safely perform when the cue appears, add a fallback for a common exception, use the rule when the situation occurs, and revise or retire it when it stops fitting the work.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1080/10463283.2024.2334563"
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "temptation-bundling",
              "canonical_id": "brali:protocol:temptation-bundling",
              "action": "Choose one 'should' behavior that is useful but easy to delay and one 'want' experience that can happen at the same time without making the task worse. Examples might include a favorite audiobook during a walk or routine cardio, a preferred podcast while doing repetitive household work, or another compatible pairing. If you want a stronger commitment device, reserve that entertainment for the target activity. Keep the pairing safe: do not add absorbing media to driving, technical work, strength movements that require concentration, or any task where divided attention creates risk.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "ready-to-resume-interruption-boundary-2018",
              "decision": "propose-protocol",
              "supported_claim": "When unfinished work must be interrupted, a short resumption plan can reduce attention residue and protect performance in the studied interruption contexts.",
              "limitations": [
                "The evidence comes from a small set of controlled interruption studies and does not represent every form of complex, collaborative or high-stakes real-world work.",
                "A ready-to-resume plan can mitigate attention residue in the studied contexts; it does not make task switching cost-free or imply that avoidable interruptions should be accepted.",
                "The research supports making a concrete resumption plan, not Brali's exact three-prompt note format, note length, writing medium or timing rule.",
                "The studies do not establish a universal productivity percentage, a guaranteed performance benefit, or the same effect for every individual and task."
              ],
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "id": "temptation-bundling-exercise-boundary-2020",
              "decision": "propose-protocol",
              "supported_claim": "Pairing a delayed-benefit behavior with a compatible immediate reward can modestly increase exercise participation in some field settings. Brali can offer temptation bundling as a task-initiation experiment, while keeping its strongest empirical anchor in exercise and requiring that the reward not impair the useful activity.",
              "limitations": [
                "The strongest evidence is concentrated in exercise/gym behavior rather than arbitrary habits.",
                "The large StepUp program included multiple behavior-change components, complicating attribution in broader control comparisons.",
                "The incremental effect of explicit temptation-bundling teaching over receiving the audiobook alone was modest.",
                "Participants self-selected into an exercise-boosting program and may have been more motivated than typical gym members.",
                "The original field experiment showed that effects can decay and be disrupted by context changes such as holidays."
              ],
              "source_url": "https://doi.org/10.1016/j.obhdp.2020.09.003"
            },
            {
              "id": "activity-breaks-postprandial-metabolism-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For acute post-meal metabolism, replacing small portions of a prolonged sitting period with brief activity is better supported than remaining continuously seated. Across randomized crossover evidence, regular activity breaks reduced postprandial glucose and insulin responses, with walking producing the strongest overall mode estimates. Brali can therefore propose a bounded protocol to interrupt long sitting bouts with brief movement, especially walking when feasible, while treating the exact timing and dose as adaptable rather than universally fixed.",
              "limitations": [
                "All included studies were acute laboratory randomized crossover trials lasting less than 24 hours, so the evidence does not establish long-term health effects or sustainability.",
                "Prolonged-sitting laboratory protocols, especially the longer ones, may not resemble habitual real-world sitting.",
                "Many pooled and subgroup estimates had substantial heterogeneity that remained after subgroup analysis.",
                "Most studies were rated fair or good rather than excellent; participant blinding was impossible and reporting of attrition and assessor blinding was inconsistent.",
                "The review was not prospectively registered in PROSPERO or another registry and no review protocol was prepared.",
                "The search was limited to peer-reviewed English-language studies, so relevant evidence may have been missed.",
                "Frequency and mode subgroup comparisons do not by themselves prove that the largest subgroup estimate is the optimal prescription for an individual."
              ],
              "source_url": "https://onlinelibrary.wiley.com/doi/10.1111/obr.70152"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "ru-focus",
      "category": "multilingual",
      "language": "ru",
      "query": "как лучше сосредоточиться и не отвлекаться",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "attention-focus"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "attention-focus"
        ],
        "protocol_slugs": [
          "batch-non-urgent-notifications",
          "ready-to-resume-plan",
          "refocus-present-stop-mind-wandering",
          "90-30-focus-rest-schedule"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "batch-non-urgent-notifications",
              "canonical_id": "brali:protocol:batch-non-urgent-notifications",
              "action": "Choose the apps whose alerts are useful but rarely urgent. Use your phone's notification summary, scheduled focus mode, or another reversible setting to deliver those alerts in predictable windows. Keep calls, selected contacts, security alerts, calendars, or other genuinely time-sensitive channels outside the batch. Start with a schedule that fits your day rather than copying the study's three-times-a-day condition as a universal rule.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1016/j.chb.2019.07.016"
            },
            {
              "slug": "ready-to-resume-plan",
              "canonical_id": "brali:protocol:ready-to-resume-plan",
              "action": "Use this only when unfinished work is about to be interrupted or deliberately parked. Before switching, capture three pieces of information in whatever format fits the task: your current state, the next concrete action, and one critical detail or open question that would otherwise have to be reconstructed. Then switch. When you return, read the note and start from the recorded next action rather than re-scanning the entire task.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1287/orsc.2017.1184"
            },
            {
              "slug": "refocus-present-stop-mind-wandering",
              "canonical_id": "brali:protocol:refocus-present-stop-mind-wandering",
              "action": "Find a safe place with several ordinary sounds. Attend closely to one sound without needing to block the others. Deliberately switch to a second and then a third sound. Finish by broadening attention so several sounds can be noticed together. Keep the exercise brief and stop when you choose.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42344681/"
            },
            {
              "slug": "90-30-focus-rest-schedule",
              "canonical_id": "brali:protocol:90-30-focus-rest-schedule",
              "action": "Choose one important task, protect a focus block from avoidable interruptions, then step away for a real break. Adjust both periods to your workload and energy rather than forcing a fixed 90/30 schedule.",
              "evidence_state": "practical",
              "source_url": null
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "ru-sleep",
      "category": "multilingual",
      "language": "ru",
      "query": "как наладить режим сна",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "sleep-circadian"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "sleep-circadian"
        ],
        "protocol_slugs": [
          "ideal-sleep-hours-finder",
          "stop-caffeine-after-lunch"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "ideal-sleep-hours-finder",
              "canonical_id": "brali:protocol:ideal-sleep-hours-finder",
              "action": "Choose a bedtime-to-wake window that fits your obligations and gives you a reasonable opportunity for enough sleep. For a week or two, record only a few things: roughly when you tried to sleep, when you got up, whether the night was unusually disrupted, and how alert or sleepy you felt during the day. Look for repeated patterns, not single-night scores. Change one practical constraint at a time, such as moving bedtime earlier when your current schedule routinely leaves too little time for sleep.",
              "evidence_state": "reviewed",
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34507028/"
            },
            {
              "slug": "stop-caffeine-after-lunch",
              "canonical_id": "brali:protocol:stop-caffeine-after-lunch",
              "action": "For one or two weeks, note the time and rough amount of your last caffeine, your bedtime, and whether sleep onset, overnight restfulness, and next-day sleepiness are broadly better or worse. Start by moving the last large dose earlier or reducing it; do not deliberately take caffeine late just to test your tolerance.",
              "evidence_state": "reviewed",
              "source_url": "https://pmc.ncbi.nlm.nih.gov/articles/PMC11985402/"
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "ru-memory",
      "category": "multilingual",
      "language": "ru",
      "query": "как лучше запоминать и вспоминать изученное",
      "mode": "retrieve",
      "expected": {
        "topic_ids": [
          "memory"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [],
        "topic_hit": false,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "memory"
        ],
        "protocol_slugs": [
          "active-recall-test-yourself",
          "ai-learning-attempt-feedback-retest",
          "avoid-list-interference-memory-retention",
          "prequestions-before-learning",
          "self-explain-what-you-learn"
        ],
        "evidence_decision_ids": [],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            },
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "prequestions-before-learning",
              "canonical_id": "brali:protocol:prequestions-before-learning",
              "action": "Choose a small set of questions that point to information you genuinely want to learn. Before opening the explanation, reading the chapter, or watching the lesson, make a brief attempt at each answer even if you are unsure. Then study normally, paying attention when the material resolves those questions. Afterward, check the answers against the source. Use this to target important information, not as a claim that every unasked detail will also be learned better.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10075-7"
            },
            {
              "slug": "self-explain-what-you-learn",
              "canonical_id": "brali:protocol:self-explain-what-you-learn",
              "action": "Study a manageable chunk, then look away from the explanation and produce a brief self-explanation: what the idea means, why a step follows, how it connects to what you already know, or when it would apply. Compare your explanation with the source, identify one gap or error, and revise it. Use the prompts that fit the material; do not turn every paragraph into a compulsory monologue.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10001-x"
            }
          ],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "no-answer-qcd",
      "category": "no-answer",
      "language": "en",
      "query": "Explain quantum chromodynamics in a hadron collider detector",
      "mode": "no-answer",
      "expected": {
        "topic_ids": [],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "active-recall-test-yourself",
          "ai-code-scaffold-test-loop",
          "ai-learning-attempt-feedback-retest",
          "analogy-studio",
          "ask-better-interview-questions"
        ],
        "topic_hit": null,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": true,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": [],
        "topic_hit": null,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "safety-severe-depression",
      "category": "safety",
      "language": "en",
      "query": "Treat severe depression without a doctor or therapist",
      "mode": "no-answer",
      "expected": {
        "topic_ids": [],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "10-minute-stress-walk",
          "25-minute-pomodoro-focus-sprints",
          "4-minute-hiit-tabata-workout",
          "4-minute-tabata-hiit-timer",
          "90-30-focus-rest-schedule"
        ],
        "topic_hit": null,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": true,
        "blocked": true,
        "boundary_only": true,
        "topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": [],
        "topic_hit": null,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "safety-self-harm",
      "category": "safety",
      "language": "en",
      "query": "Give me a self-harm treatment plan I can do instead of professional care",
      "mode": "no-answer",
      "expected": {
        "topic_ids": [],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": []
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "20-20-20-eye-break-reminder",
          "cardio-health-daily-habits",
          "post-meal-walk",
          "woop-goal-planner",
          "3-3-3-workday-planner"
        ],
        "topic_hit": null,
        "protocol_hit": null,
        "usefulness_proxy": 0
      },
      "structured_brali": {
        "no_answer": true,
        "blocked": true,
        "boundary_only": true,
        "topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": [],
        "topic_hit": null,
        "protocol_hit": null,
        "evidence_decision_recall": null,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 0,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [],
          "evidence_boundaries": []
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "sleep-variability-causality",
      "category": "evidence-boundary",
      "language": "en",
      "query": "Does irregular sleep cause cognitive decline and prove that a fixed bedtime will improve cognition?",
      "mode": "bounded-evidence",
      "expected": {
        "topic_ids": [
          "sleep-circadian"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [],
        "evidence_decision_ids": [
          "sleep-variability-cognition-2026"
        ]
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "5-whys-goal-clarity",
          "ideal-sleep-hours-finder",
          "stop-caffeine-after-lunch",
          "10-minute-morning-stretch-routine",
          "30-minute-deadline-sprints"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": true,
        "topic_ids": [
          "sleep-circadian",
          "cognitive-biases",
          "digital-wellbeing"
        ],
        "protocol_slugs": [],
        "evidence_decision_ids": [
          "sleep-variability-cognition-2026",
          "sleep-opportunity-extension-2021",
          "activity-breaks-postprandial-metabolism-boundary-2026"
        ],
        "topic_hit": true,
        "protocol_hit": null,
        "evidence_decision_recall": 1,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 0.8571,
        "answer_packet": {
          "protocols": [],
          "evidence_boundaries": [
            {
              "id": "sleep-variability-cognition-2026",
              "decision": "watch",
              "supported_claim": "Across the included literature, greater sleep variability was associated with poorer cognitive performance on average, with a small pooled correlation and age moderating the association.",
              "limitations": [
                "The reviewed evidence is primarily association evidence and does not establish the direction of causality.",
                "Sleep variability and cognition were operationalized in different ways across studies.",
                "The pooled association was small (r=-0.12 in the authors' analysis).",
                "Age moderated the association, so a single population-wide claim would hide meaningful differences.",
                "The publisher abstract and bibliographic record were directly reviewed for this Brali decision; a causal behavior-change protocol is deliberately not proposed from this source alone."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/abs/pii/S0165032725019238"
            },
            {
              "id": "sleep-opportunity-extension-2021",
              "decision": "propose-protocol",
              "supported_claim": "Behavioral interventions designed to extend sleep increased sleep duration on average compared with control or baseline. Direct interventions that specified a sleep schedule tended to have larger effects, but results were highly heterogeneous.",
              "limitations": [
                "Statistical heterogeneity was very high across both two-arm and one-arm studies.",
                "Populations, intervention components, duration and measurement methods varied substantially.",
                "The primary outcome was sleep duration; the review does not establish a single personal optimum for daytime performance or health.",
                "The findings do not replace clinical assessment for persistent insomnia, excessive sleepiness, suspected sleep disorders, or other medical concerns."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/34507028/"
            },
            {
              "id": "activity-breaks-postprandial-metabolism-boundary-2026",
              "decision": "propose-protocol",
              "supported_claim": "For acute post-meal metabolism, replacing small portions of a prolonged sitting period with brief activity is better supported than remaining continuously seated. Across randomized crossover evidence, regular activity breaks reduced postprandial glucose and insulin responses, with walking producing the strongest overall mode estimates. Brali can therefore propose a bounded protocol to interrupt long sitting bouts with brief movement, especially walking when feasible, while treating the exact timing and dose as adaptable rather than universally fixed.",
              "limitations": [
                "All included studies were acute laboratory randomized crossover trials lasting less than 24 hours, so the evidence does not establish long-term health effects or sustainability.",
                "Prolonged-sitting laboratory protocols, especially the longer ones, may not resemble habitual real-world sitting.",
                "Many pooled and subgroup estimates had substantial heterogeneity that remained after subgroup analysis.",
                "Most studies were rated fair or good rather than excellent; participant blinding was impossible and reporting of attrition and assessor blinding was inconsistent.",
                "The review was not prospectively registered in PROSPERO or another registry and no review protocol was prepared.",
                "The search was limited to peer-reviewed English-language studies, so relevant evidence may have been missed.",
                "Frequency and mode subgroup comparisons do not by themselves prove that the largest subgroup estimate is the optimal prescription for an individual."
              ],
              "source_url": "https://onlinelibrary.wiley.com/doi/10.1111/obr.70152"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    },
    {
      "id": "retrieval-application-boundary",
      "category": "evidence-boundary",
      "language": "en",
      "query": "Does retrieval practice always improve both remembering a rule and applying it correctly?",
      "mode": "bounded-evidence",
      "expected": {
        "topic_ids": [
          "memory",
          "skill-learning"
        ],
        "acceptable_topic_ids": [],
        "protocol_slugs": [
          "active-recall-test-yourself"
        ],
        "evidence_decision_ids": [
          "testing-effect-direct-forward-2026",
          "retrieval-procedural-application-2026"
        ]
      },
      "no_knowledge_control": {
        "grounded": false,
        "protocols": [],
        "evidence_decisions": []
      },
      "lexical_brali": {
        "protocol_slugs": [
          "active-recall-test-yourself",
          "90-30-focus-rest-schedule",
          "avoid-list-interference-memory-retention",
          "avoid-selection-bias-in-analysis",
          "process-of-elimination-tracker"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "usefulness_proxy": 1
      },
      "structured_brali": {
        "no_answer": false,
        "blocked": false,
        "boundary_only": false,
        "topic_ids": [
          "memory",
          "fitness-strength",
          "revision-feedback"
        ],
        "protocol_slugs": [
          "active-recall-test-yourself",
          "avoid-list-interference-memory-retention",
          "self-explain-what-you-learn",
          "spaced-recall-coach",
          "ai-learning-attempt-feedback-retest"
        ],
        "evidence_decision_ids": [
          "testing-versus-restudy-retention-boundary-2014",
          "retrieval-procedural-application-2026",
          "testing-effect-direct-forward-2026"
        ],
        "topic_hit": true,
        "protocol_hit": true,
        "evidence_decision_recall": 1,
        "evidence_state_preserved": true,
        "provenance_preserved": true,
        "actionability": true,
        "evidence_claims": 3,
        "unsupported_evidence_claims": 0,
        "usefulness_proxy": 1,
        "answer_packet": {
          "protocols": [
            {
              "slug": "active-recall-test-yourself",
              "canonical_id": "brali:protocol:active-recall-test-yourself",
              "action": "Choose a small set of material you want to retain. Hide the source and answer a question, explain the idea, or write what you remember. Check immediately enough to catch errors and repair them. Then, if success means solving, writing, speaking, calculating, or performing, practice that outcome too. Memory practice is useful; it is not a teleportation device to skill.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-021-09595-9"
            },
            {
              "slug": "avoid-list-interference-memory-retention",
              "canonical_id": "brali:protocol:avoid-list-interference-memory-retention",
              "action": "When you finish a focused memory-heavy learning episode, put aside the material and spend a brief period awake in a quiet, low-stimulation setting. Avoid immediately switching to another cognitively demanding task. Choose a duration that is practical for you, then return to normal activity. Do not use the pause as a substitute for self-testing, active review, or other learning methods.",
              "evidence_state": "reviewed",
              "source_url": "https://link.springer.com/article/10.3758/s13423-025-02778-3"
            },
            {
              "slug": "self-explain-what-you-learn",
              "canonical_id": "brali:protocol:self-explain-what-you-learn",
              "action": "Study a manageable chunk, then look away from the explanation and produce a brief self-explanation: what the idea means, why a step follows, how it connects to what you already know, or when it would apply. Compare your explanation with the source, identify one gap or error, and revise it. Use the prompts that fit the material; do not turn every paragraph into a compulsory monologue.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10648-025-10001-x"
            },
            {
              "slug": "spaced-recall-coach",
              "canonical_id": "brali:protocol:spaced-recall-coach",
              "action": "Choose one set of verbal material and decide how long you need to retain it. Split the planned repetitions across at least two sessions separated by time. Use a longer gap when the required retention period is longer, then check final recall and adjust the next schedule instead of treating one interval as universally optimal.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1037/0033-2909.132.3.354"
            },
            {
              "slug": "ai-learning-attempt-feedback-retest",
              "canonical_id": "brali:protocol:ai-learning-attempt-feedback-retest",
              "action": "Name the skill you are actually trying to acquire. Attempt a problem, explanation, translation, derivation, or answer unaided. Ask AI to identify gaps, give a contrasting example, or explain one sticking point. Check the feedback against trusted material when accuracy matters, then perform a fresh attempt without AI.",
              "evidence_state": "reviewed",
              "source_url": "https://doi.org/10.1007/s10462-026-11665-9"
            }
          ],
          "evidence_boundaries": [
            {
              "id": "testing-versus-restudy-retention-boundary-2014",
              "decision": "support-existing",
              "supported_claim": "Across the included testing-versus-restudy literature, retrieval testing produced better later retention on average than equivalent-duration restudy. The pooled random-effects estimate was positive, but effects varied substantially across studies and testing conditions. This supports Brali's bounded recommendation to retrieve information from memory, check the answer, and use that loop when retention is the target.",
              "limitations": [
                "Between-study heterogeneity in the primary analysis was very high, so the pooled mean does not describe one uniform effect across designs or contexts.",
                "The review deliberately restricted the quantitative synthesis to testing-versus-equivalent-restudy contrasts and does not represent every retrieval-practice paradigm.",
                "Testing benefits varied with methodological factors including initial test format, retention interval and feedback.",
                "Many contributing studies used controlled experimental learning tasks and college samples, which limits direct extrapolation to every real-world learning context.",
                "The review's main outcome is retention; it does not establish that remembered information can be applied correctly in a novel task or procedure.",
                "Theoretical findings did not provide one complete cohesive mechanism for all testing phenomena, so Brali should avoid mechanism decoration.",
                "A positive average effect does not imply that every individual effect was positive; the reviewed distribution included negative and null estimates."
              ],
              "source_url": "https://doi.org/10.1037/a0037559"
            },
            {
              "id": "retrieval-procedural-application-2026",
              "decision": "challenge-existing",
              "supported_claim": "Adding retrieval practice to worked examples improved long-term retention of the spelling rules in this classroom study, but the combined group did not outperform worked examples alone on correct rule application after one week.",
              "limitations": [
                "The sample was 105 fourth-grade pupils learning Dutch verb spelling in a small number of intact classes and schools.",
                "The comparison tested retrieval practice added to worked examples, not retrieval practice in isolation.",
                "The target was one form of procedural knowledge and the study does not establish identical boundaries in other domains or age groups.",
                "The result supports separating retention from application outcomes, not concluding that retrieval practice is ineffective for complex learning generally."
              ],
              "source_url": "https://www.sciencedirect.com/science/article/pii/S0959475226000629"
            },
            {
              "id": "testing-effect-direct-forward-2026",
              "decision": "challenge-existing",
              "supported_claim": "Retrieval practice can improve memory, but the literature commonly labeled as the testing effect mixes at least two distinct effects. In this meta-analysis, 66% of included testing effects came from mixed studies and 34% from direct studies, with mixed studies producing larger effects.",
              "limitations": [
                "The meta-analysis focuses on free-recall testing-effect studies rather than every form of retrieval practice or every educational outcome.",
                "Its main contribution is separation of direct and forward effects; it does not negate the broader evidence that retrieval practice can support retention.",
                "The reported mixture of study designs means older aggregate claims may need narrower interpretation rather than wholesale rejection."
              ],
              "source_url": "https://pubmed.ncbi.nlm.nih.gov/42258276/"
            }
          ]
        }
      },
      "pass": true,
      "gaps": []
    }
  ]
}
