skills.rest
lfuuu
lfuuuCOMMUNITY
Explore all skills by this author →
GitHub

repo-audit

Audits a repository for architecture, testing, security, and documentation issues with ranked findings.

Updated Jun 2, 2026
Search Tags:#code quality#technical debt#security review#repository audit#todo backlog#subagent fan-out#findings ledger
One-click install
npx skills add https://github.com/lfuuu/claude-rules --skill repo-audit-lfuuu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: repo-audit
Source: https://github.com/lfuuu/claude-rules/tree/main/global-skills/repo-audit
Command: npx skills add https://github.com/lfuuu/claude-rules --skill repo-audit-lfuuu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? It automates whole-repository audits, producing evidence-backed findings (with file:line references, concrete fixes, and impact/effort estimates) instead of vague impressions of technical debt. ## Core Features & Use Cases - Four audit dimensions: architecture/maintainability, tests/reliability, security, and documentation/DX/CI, each with defined signals and delegate skills. - Quick and deep modes: inline high-confidence scans by default, or fan-out subagent analysis for deeper coverage. - Deduplication ledger: a versioned ledger in docs/audit prevents already-filed or rejected findings from resurfacing in later runs. - Approval-gated backlog filing: ranked findings are presented for approval, and only approved items are filed as /todo drafts. - Use Case: Ask for a deep audit of a legacy service; receive a ranked report of critical security and testing gaps, approve three items, and have them filed as todo drafts with evidence. ## Quick Start Run repo-audit in deep mode on this repository and show me the ranked findings.

Frequently Asked Questions about repo-audit

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit a repository for technical debt and code problems?▼

Run repo-audit, which scans four dimensions (architecture, tests, security, documentation/DX) and returns a ranked report. Each finding includes file:line evidence, a concrete fix, and impact/effort estimates for prioritization.

What is the difference between quick and deep audit modes?▼

Quick mode is the default and runs an inline scan collecting high-confidence findings cheaply. Deep mode fans out subagents per dimension and area for broader, deeper analysis at higher cost.

Does repo-audit create backlog items automatically?▼

No. It presents a ranked list and waits for explicit approval before filing anything. Approved findings are filed as /todo drafts; rejected ones are recorded in the ledger so they never resurface.

How does the audit avoid repeating findings from previous runs?▼

A versioned ledger at docs/audit/repo-audit-ledger.md stores stable finding ids marked as filed or rejected. Findings matching those ids are dropped during deduplication in subsequent runs.

Is repo-audit tied to a specific programming language or stack?▼

No. It is stack-agnostic: it detects the project stack and layout during reconnaissance, uses existing docs-system documentation as context when available, and delegates specialized analysis to dedicated skills.

Related Skills

Explore More Software Engineering Skills→
robotics-playgroundrobotics-playground

bmad-document-project

Generate index, architecture, data models, API contracts, and deployment notes for brownfield codebases.

Official
Advanced
rethinkhealthrethinkhealth
6

todo-triage

Automate triage of pending todos with severity, category, and recommended action.

Official
Advanced
SwitchbackTechSwitchbackTech
228

improve-codebase-architecture

Identify shallow modules and cross-cutting seams in modular or monorepo codebases.

Official
Advanced
Cogni-AI-OUCogni-AI-OU
3

security-review

Analyze pull request diffs to detect security vulnerabilities in incremental changes.

Official
Basic