agentleFS
tyler-berggrentyler-berggren/tb-claude-kit/.claude/skills/vibe-audit/SKILL.md

vibe-audit

Skill2 starsChanged 5 days agoMITView on GitHub

Codebase health + security audit for vibe-engineered projects. Scans code, checks architecture guardrails, reviews threat model. Findings brainstormed with user before committing to brain.

6 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read cowork/vibe-audit/PROCEDURE.md for the full audit procedure — three phases (scan, brainstorm, commit), scan categories, VIBE-AUDIT.db …

    line 24
  2. Very specific:

    Read cowork/vibe-audit/GUARDRAILS.md for the Layer 2 architecture guardrails checklist.

    line 26
  3. Very specific:

    If .claude/kit.json has a rules."vibe-audit" entry, read it and apply it as an additional instruction for this skill.

    line 53

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Last Vibe Audit

!sqlite3 -separator ' | ' cowork/brain/BRAIN.db "SELECT '#' || id, created_at, substr(body, 1, 150) FROM logs WHERE type = 'milestone' AND tags LIKE '%vibe-audit%' ORDER BY created_at DESC LIMIT 1;" 2>/dev/null || echo "No prior vibe-audit found"

Pattern Effectiveness

!sqlite3 -separator ' | ' cowork/vibe-audit/VIBE-AUDIT.db "SELECT slug, tp || 'tp/' || fp || 'fp', effectiveness_grade FROM v_pattern_effectiveness WHERE tp + fp > 0 ORDER BY precision DESC LIMIT 10;" 2>/dev/null || echo "No pattern data yet — first run will establish baseline"

Last 3 Runs

!sqlite3 -separator ' | ' cowork/vibe-audit/VIBE-AUDIT.db "SELECT '#' || id, started_at, scope, summary FROM runs ORDER BY started_at DESC LIMIT 3;" 2>/dev/null || echo "No prior runs"

Open Audit Findings

!sqlite3 -separator ' | ' cowork/brain/BRAIN.db "SELECT '#' || id, type, title FROM logs WHERE tags LIKE '%vibe-audit%' AND type IN ('task', 'question') AND status = 'active' ORDER BY importance DESC, created_at DESC LIMIT 10;" 2>/dev/null || echo "None"

Threat Model Entries

!sqlite3 -separator ' | ' cowork/brain/BRAIN.db "SELECT '#' || id, title FROM logs WHERE tags LIKE '%threat-model%' AND status = 'active' ORDER BY importance DESC;" 2>/dev/null || echo "None"


Read cowork/vibe-audit/PROCEDURE.md for the full audit procedure — three phases (scan, brainstorm, commit), scan categories, VIBE-AUDIT.db integration.

Read cowork/vibe-audit/GUARDRAILS.md for the Layer 2 architecture guardrails checklist.

Key paths

  • VIBE-AUDIT.db: cowork/vibe-audit/VIBE-AUDIT.db — self-learning audit database with Bayesian pattern tracking
  • Schema: cowork/vibe-audit/schema.sql — recreate DB if missing
  • Seed data: cowork/vibe-audit/seed.sql — builtin patterns and suppressions
  • Procedure: cowork/vibe-audit/PROCEDURE.md — full scan/brainstorm/commit procedure
  • Guardrails: cowork/vibe-audit/GUARDRAILS.md — Layer 2 architecture checklist

Rules

  • No permission needed for the scan — Phase 1 executes immediately.
  • Always brainstorm before committing — Phase 2 is mandatory. Never write findings to the brain without user discussion and confirmation.
  • Speed over perfection — The scan is a quick health check, not a penetration test. Use grep/wc/bash. Target under 90 seconds for Phase 1.
  • Be smart about intentional patterns — Read surrounding context before flagging. Don't flag patterns that are explicitly mitigated.
  • Severity must be justified — Critical = actively dangerous or harmful to productivity. Warning = address soon. Info = worth noting, no action yet.
  • Concrete recommendations — Every Critical/Warning finding needs a specific, actionable fix (not "consider improving").
  • No false courage — If you can't determine whether a pattern is safe, flag it as Warning with a note for manual review.
  • Normalize tags — Split by comma, trim, lowercase, sort alphabetically, deduplicate, rejoin.
  • The brain is the report — Findings are brain entries. The milestone is the summary. VIBE-AUDIT.db tracks scan mechanics and learning — it's the skill's memory, not the project's knowledge base.
  • VIBE-AUDIT.db is the skill's memory — Every run creates a record, every finding gets a fingerprint and disposition. Patterns learn from feedback. Suppressions accumulate. The skill gets smarter each run.

Project overrides

If .claude/kit.json has a rules."vibe-audit" entry, read it and apply it as an additional instruction for this skill. Absent file or key means no overrides — that is the normal case.

jq -r '.rules."vibe-audit" // empty' .claude/kit.json 2>/dev/null

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.