Managed MCP Doctor — Security skill for Claude Code
Audit Claude Code MCP governance drift, catch when allowedMcpServers no longer covers managedMcpServers (2.1.259+).
How to install Managed MCP Doctor
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open Onur45500/managed-mcp-doctor and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Managed MCP Doctor does
Audit Claude Code MCP governance drift, catch when allowedMcpServers no longer covers managedMcpServers (2.1.259+). CLI + JSON/SARIF for CI.
Alternatives in Security
- Node9 Proxy — The Execution Security Layer for the Agentic Era 209 ★
- AWS MCP Server — by alexei-led - Features multiple Python environment setup options with detailed code style guidelines, compre 182 ★
- Sovereign Skills — 20 production-grade skills for AI coding agents — setup, scope, discipline, code review, security, session man 125 ★
README
managed-mcp-doctor
[](https://github.com/Onur45500/managed-mcp-doctor/actions/workflows/ci.yml) [](https://www.npmjs.com/package/managed-mcp-doctor) [](https://nodejs.org) [](LICENSE)
Your `allowedMcpServers` allowlist stopped covering org-pushed MCP servers in Claude Code **2.1.259**. This tells you which ones just became reachable — and can write a drop-in fix fragment.
Install & run (10 seconds)
npx managed-mcp-doctorJSON / SARIF for CI:
npx managed-mcp-doctor --json --fail-on critical
npx managed-mcp-doctor --sarif --fail-on warn > mcp-doctor.sarifGuided fix (dry-run, then write):
npx managed-mcp-doctor --fix
npx managed-mcp-doctor --fix --write --yes # writes managed-settings.d/90-mcp-doctor.jsonUnder `managedSourcesBehavior: "first-wins"`, if MDM / HKLM / remote is the winning source, `--fix --write` **refuses** (exit 2) so you do not leave a dead drop-in. Dry-run still prints the fragment with a deploy-into-winner note. Use `--force-write` only when fleet tooling syncs that JSON elsewhere. After a successful write, the tool **re-audits** and reports which finding ids were resolved vs remaining.
Sample report
Before (policy drift)
Allowlist present, no denylist, managed servers pushed — `CRITICAL` for ungoverned `managedMcpServers` entries plus related `WARN`s.
After (healthy locks)
Denylist covers managed servers, allowlist locked with `allowManagedMcpServersOnly`, `d
Related Skills
Audit MCP Mock Drift
Catch mock/live divergence in MCP eval golden tests (anti-pattern #2 from eval-mock-mode-discipline)
Raindrop Collection Governance
Agent Skills-compatible skill: govern a Raindrop.io library's collection structure via raindrop-mcp — audit, c
Sentience Governor
Local-first runtime governance for AI agents. Declare intent, catch out-of-scope actions, and audit them in a
Repo Doctor
Audit a repo's agent docs — AGENTS.md/CLAUDE.md topology, always-loaded token cost, CodeMap freshness and line
Patchbot
Vulnerability scanning with your own scanners and threat feeds, plus a coding agent that opens the fix PR. CLI
Securevector AI Threat Monitor
Security & Observability for AI Agents — audit every tool call, enforce allow/block policies, catch prompt inj
Related Agents
Docs Drift Checker
Use to detect places where code has diverged from its documentation — function signatures vs docstrings, READM
Core Owner
Catch-all owner for claude-code-vault's core modules — MCP server, indexer (chunks/embeddings/graph), CLI, and
Bh Blind Spot
Blind-spot lens of Black Heron audit. Runs LAST, after code_quality + governance + drift. Hunts meta-intersect
