Drop into a skill folder
Personal: ~/.claude/skills/clustering-illusion-guard/SKILL.md
Project: .claude/skills/clustering-illusion-guard/SKILL.md
Brali Agent Skill · practical · usable
When a short streak looks meaningful, record the complete observation window and denominator, compare the cluster with the broader history, check whether measurement or operating conditions changed, and avoid changing a plan solely because the sequence looks striking.
Portable instruction
Record what happened, the full period or set you are looking at, and the denominator behind the visible cluster. Compare it with the earlier history when available, note any change in measurement or operating conditions, identify plausible alternative explanations, and choose whether to act, gather more information or keep the current plan.
What streak caught my attention, what is the full denominator and observation window, what changed in the context, and would I make the same decision if the sequence looked less visually striking?
Install
Inspect SKILL.md first. The evidence state and skill mode are part of the contract.
Install for normal use.
Personal: ~/.claude/skills/clustering-illusion-guard/SKILL.md
Project: .claude/skills/clustering-illusion-guard/SKILL.md
hermes skills install https://brali-lifeos.github.io/skill-packs/clustering-illusion-guard/SKILL.mdWorkspace: <workspace>/skills/clustering-illusion-guard/SKILL.md
Global: ~/.openclaw/skills/clustering-illusion-guard/SKILL.md
Brali verification
Canonical ID: brali:clustering-illusion-guard
Evidence state: practical
Skill mode: usable
Recommendation eligible: yes
SHA-256: 3a63c5f29a30651c2ec040a5e5a93360efc151354b652dd5f2477d610c327516
Brali CI verified. This means identity, state, mode, provenance and file integrity matched the generated corpus. It is not third-party vendor endorsement.
Brali keeps a one-to-one machine surface: one hack record, one stable skill identity. Trust state determines what the skill is allowed to do.
Original Brali knowledge is available for non-commercial reuse under CC BY-NC-SA 4.0. Keep attribution, canonical record link, evidence state and skill mode.