codebase-audit 开源项目推荐（19 个）

levnikolaevich /claude-code-skills

Claude Code的插件套件+捆绑的LCP服务器。完整的交付生命周期：具有多模型人工智能审查、项目引导、文档生成、代码库审计、性能优化、社区工作流程的敏捷管道。包括hex-line（哈希验证编辑）、hex-graph（代码知识图）和hex-ssh（远程SSH）HCP服务器。

eddiebelaval /squire

The most complete Claude Code toolkit in the wild. 348 skills, 80 commands, 35 agents, stage-gate pipelines, multi-model orchestration, and thinking frameworks -- distilled from six months and 19,030 commits across 104 repositories.

Dev-Wiki /dev-harness

Engineering guardrails and reusable skills for AI coding agents: project context, planning, documentation, verification, Git workflows, bug fixes, and codebase audits.

sturec5 /code-audit-protocol

LLM-assisted code auditing. 54 phases, custom overlays, accountability.

flavorgood /Selective-Intelligence

Selective Intelligence turns minimal intent into grounded, token-efficient AI work: research, product design, one-prompt websites, verified UI/UX, repository realignment, and developer-grade execution across AI clients—free, open, and model-neutral.

Calvin-LLC /agentic-hardening-skill

Evidence-driven codebase hardening skill. Audits OWASP Top 10:2025, OWASP LLM and Agentic Top 10 when AI is detected, supply chain (SBOM, SLSA v1.2), reliability, and tests. Reports only. Agent Skills layout.

welshofer /elevate

A Claude skill that audits a codebase end-to-end and produces a top-5-per-dimension improvement plan across seven lenses (Performance, Functionality, Stability, Reliability, Security, Usability, Attractiveness).

kartikkabadi /grok-brownfield

Grok Build skill: audit, validate, and improve existing (brownfield) codebases with assumption-aware analysis and optional PR execution

Sayista-Yazdani /Gemma-Doc-Pro

Gemma Doc Pro is a privacy-first, serverless codebase auditing and interactive documentation platform.

umutgungorr /envdoctor

🩺 Zero-dependency .env & .env.example linter, synchronizer, and code auditor CLI. Prevents missing environment variables and deploy crashes.

IoannisBekas /Y.Skill

Self-improving repo health remediation skill with audit/fix/diff modes, evidence coverage, counter-review, and stable HP-* findings.

tinosingh /Code-Health-Check-Prompt

Codebase Health Audit Prompt

ydpzg /repo-ritual

A fast, local-first repository health check and onboarding brief with strict CI gates and shareable reports.

faridjafarlee /recon

Multi-phase codebase recon → improve for Claude Code: map → hunt → gated fix/optimize, plus feature suggest/plan. Installable plugin.

sowadalmughni /ai-codebase-audit

Audits AI-generated codebases for the six failure modes that cause production collapse: disconnected schema, missing RLS, exposed secrets, broken auth, unwired integrations, and N+1 queries. Outputs a severity-ranked remediation register, not inline fixes.

batu3384 /codebase-audit

Multi-host Agent Skill — evidence-backed whole-repo architecture and maintainability audit. Cursor, Claude Code, Codex, Antigravity. macOS/Linux/Windows.

agiphi /veridence-sovereign-engine

Veridence: Deterministic Algorithmic Technical Due Diligence (The Sovereignty Protocol & 4-Quadrant Infrastructure Risk Quantification). Aligned with the August 2026 Technical Report.

krwhynot /toque

Toque: AI-powered codebase grading, planning, auditing, and operational readiness. Claude Code plugin.

Petrcatapultian539 /Code-Health-Check-Prompt

Audit code health with AI using evidence-based checks, risk metrics, and prioritized findings for clear refactor decisions