# mcp-audit-tool

> MIT static audit of MCP client configs — poisoning, rugs, secrets

**Category:** AI Agents  
**Pricing:** open-source  
**URL:** https://github.com/graygnatconsole/mcp-audit-tool  
**GitHub:** https://github.com/graygnatconsole/mcp-audit-tool  
**Added:** 2026-10-04

---

## The Hook

mcp-audit-tool is a MIT pure-Python CLI that statically audits Claude/Cursor/VS Code MCP client configs for tool poisoning, rug-pull installs, hardcoded secrets, and supply-chain smells — before the model runs them.

## What's Good

MIT (graygnatconsole/mcp-audit-tool). `pipx install git+https://github.com/graygnatconsole/mcp-audit-tool.git` then `mcp-audit scan` (auto-discovers client configs). 12 rules aimed at OWASP MCP Top 10-class issues; SARIF/JSON for CI; no LLM key required. Distinct from live MCP server probing — this is config static analysis. ★~79.

## The Catch

Package: README PyPI badge may lag — install from the MIT git repo / pipx, not the unrelated PyPI name `mcpaudit` (different project). Vs mcp-scanner / hol-guard / skillspector: MCP Scanner audits MCP servers/tools; HOL Guard is runtime local allow/block; SkillSpector gates skills — mcp-audit-tool’s catch is client-config static audit. ★ still small; pin a commit; findings are leads, not proof.

## Verdict

MIT MCP client-config auditor when you gate `.cursor/mcp.json` before connect — not Cisco MCP Scanner servers and not HOL Guard runtime policy.