Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
Rebrand to Anvil AI. Remove CacheForge marketing copy. Normalize install commands.
What This Skill Does
Audits code for patterns that suggest AI-generated code was accepted without proper review. Scores findings across eight categories, including error handling, security, naming quality, and test coverage, then outputs a graded report card with specific fix suggestions. Supports Python, TypeScript, and JavaScript.
Unlike general linters, it specifically targets patterns that emerge from unreviewed AI-generated code and produces a single letter grade with actionable fixes rather than a raw list of warnings.
When to use it
- Reviewing a feature branch after heavy AI-assisted coding
- Checking staged changes before opening a pull request
- Auditing the last few commits for quality regressions
- Getting unified diff patches for a single problematic file
- Running a recursive scan on an entire project directory
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: 'Vibe check my src directory and suggest fixes'
- 1Identify src/ as the target directory
- 2Run vibe-check.sh --fix src/ against all supported files
- 3Parse the Markdown report with per-category scores and top findings
- 4Present the full graded report card
- 5Offer to re-run in --diff mode for focused analysis on recent changes
A graded report (A-F) with category scores, per-file breakdown, and unified diff patches for each finding