Project Audit Rules

What it does

This skill runs project-specific audit rules (markdown-based) that validate business logic, architecture patterns, and naming conventions which standard linters/test suites miss. It can collect relevant changed files, run scripted checks, AI-evaluate results against pass criteria, and produce a readable audit report with PASS/FAIL/WARN classifications.

When to use it

Use this when you need an automated quality gate that enforces architectural decisions, naming conventions, or domain-specific checks during CI or code review, or to bootstrap rule sets for new projects. Invoke via /audit or /audit manage to run or manage rules.

What's included

• Scripts: No executable scripts included in this skill (has_scripts=false) but the skill documents the rule format and check workflow.
• References: Yes — the repo contains a references/ directory with examples and rule-format guidance.
• Instructions: The SKILL.md explains RUN vs MANAGE modes, how to collect rules from .claude/audit-rules/*.md, execute checks, and report results.

Compatible agents

Best used with Claude Code / SuperClaude-style Claude Code frameworks and any agent that can execute shell scripts and parse markdown-based rules (Claude Code, SpoonOS-style agents).