Does every Brali hack have an Agent Skill?

Yes. The build creates a deterministic SKILL.md, skill.json and stable skill page for every hack in the Brali corpus. New hacks receive a skill automatically on the next build.

Can every Brali skill be used as practical advice?

No. Reviewed and practical skills are usable. Pending-review skills exist for review only. Restricted skills are non-operational reference shells. Trust state is preserved inside every skill.

Can Brali skills be used with Claude Code, Hermes Agent or OpenClaw?

Yes. Brali emits portable SKILL.md files in the Agent Skills structure, with host-specific installation paths or commands for Claude Code, Hermes Agent and OpenClaw.

One hack → one Agent Skill · automatic on every build

The whole Brali hack library now has a skill layer.

Every Brali hack gets its own stable SKILL.md, machine record and skill page. When a new hack enters the corpus, its skill is generated automatically. Trust state changes what the skill is allowed to do; it never removes the skill identity.

Browse all skills Complete library.json Trusted catalog.json Trusted Protocol Feed

Use the agent you already have

One portable file, familiar agent hosts.

Claude Code

Filesystem skill

Personal: ~/.claude/skills/<skill-name>/SKILL.md
Project: .claude/skills/<skill-name>/SKILL.md

Claude Agent Skills docs →

Hermes Agent

Install directly from Brali

hermes skills install https://brali-lifeos.github.io/skill-packs/<skill>/SKILL.md

Hermes skills docs →

OpenClaw

Workspace or global root

<workspace>/skills/<skill-name>/SKILL.md
~/.openclaw/skills/<skill-name>/SKILL.md

OpenClaw skills docs →

Prefer one router?

Install one Brali router instead of hundreds of individual skills.

The repository-level brali-life-os skill searches only the trusted recommendation layer and preserves provenance. It is useful when you want the agent to select a fitting Brali protocol dynamically.

gh skill install Brali-LifeOS/brali-lifeos.github.io brali-life-os

Inspect the router skill before installing.

Complete Agent Skills marketplace

Search every Brali hack as an agent skill.

Loading the complete one-skill-per-hack library…

Search skills

Trust mode

Pomodoro focus skill AI premortem skill Feedback skill

Available skills

Select a skill

You will see its trust mode, raw SKILL.md, installation paths, canonical record and verification checksum.

Why include review-gated skills at all?

Because identity and execution are different concerns. If Brali has a hack, an agent should be able to resolve that hack to a stable skill artifact. But a file existing must never silently upgrade an unreviewed idea into advice. The skill mode is therefore generated from the evidence state and enforced in CI.

Search boundary

The marketplace itself and trusted skill pages are indexable. Review-required and restricted skill pages are generated but kept out of normal search indexing. This keeps complete machine coverage without flooding search results with material that has not met the recommendation bar. A skill page is not a ranking shortcut.

Machine-readable surfaces

library.json contains every generated skill. catalog.json is the smaller reviewed/practical recommendation catalog. Agents that need practical recommendations should use the trusted catalog or Trusted Protocol Feed rather than treating the full library as equally trustworthy.

Verification

CI checks one-skill-per-hack coverage, deterministic names, evidence-state parity, skill mode, provenance, host metadata, stable links and SHA-256 integrity. “Brali CI verified” is an integrity/provenance statement, not security certification or vendor endorsement.

Licensing and attribution

Original Brali knowledge is available for non-commercial reuse under CC BY-NC-SA 4.0. Keep Brali attribution, the canonical record link, evidence state and skill mode. Commercial use requires separate permission. See citation guidance and terms.