intelligence-review-skills — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited intelligence-review-skills (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Read-only audit of intelligence/rules/, intelligence/agents/, intelligence/skills/. Detects issues, proposes actions, hands accepted actions to intelligence-learn-from-context Phase B for application.
Name reflects the umbrella usage of "skills" for all AI artifacts (rules + skills + agents).
docs/CONVENTIONS.md Authoring Discipline section — size budgets, description discipline, writing principles. The detection checks below reference these standards.intelligence/rules/, intelligence/agents/, intelligence/skills/ (recursive — each skill is its own directory).git log -1 --format='%ai' <file>git log --oneline <file> | wc -lgit log --follow --format='%ai' <file> | tail -1Stale candidates have no recent edits and a low cross-reference count from other artifacts.
| Check | Detection | Proposed action |
|---|---|---|
| Duplicate content | Two artifacts cover overlapping scope, or descriptions share trigger phrases | MERGE — present both, propose unified |
| Over size budget | SKILL.md >1000 lines, rule >500 with multiple distinct topics | SPLIT — propose 2+ artifacts with clean scope, or extract detail to references/<topic>.md |
| Stale | git: no edits in 6+ months and zero cross-references from other artifacts | ARCHIVE — move to intelligence/_archive/ |
| Negative-framed instruction (judgment call) | Body contains "never do X" / "don't X" / "X, not Y" where positive framing fits | REWRITE — propose positive replacement |
| ALL-CAPS MUST/NEVER for judgment | Body uses uppercase MUST / NEVER / ALWAYS outside safety/security/output context | REFRAME — replace with decision rule + reasoning |
| Weak / duplicate description | Description identical or near-identical to another, or under 3 words for a sibling-rich registry | DIFFERENTIATE — propose distinct trigger |
| Missing frontmatter field | Required field absent (name, description) | PATCH — add the missing field |
| Orphan rule | No skill / agent / cross-reference points at this rule | FLAG — review whether intentional or stale |
| Body leads negative | First section of rule body is FORBIDDEN instead of REQUIRED | REORDER — REQUIRED first |
| Description over cap | Description exceeds 250 chars | TRIM — propose shorter version preserving distinct trigger |
User reviews list, accepts items individually. Bulk-accept for low-impact tweaks is fine.
intelligence-learn-from-context Phase B for application. That skill owns the write machinery — single source of truth for the apply path. Pass:MERGE / SPLIT / ARCHIVE / REWRITE / REFRAME / DIFFERENTIATE / PATCH / REORDER / TRIM)/intelligence-sync runs once for all applied changes.intelligence-learn-from-context — single-lesson capture; this skill's apply phase delegates to its Phase Bintelligence-extract-skill — when an audit surfaces a workflow that should become a new skill~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.