dr-mcp server
Audits MCP configs, scores risk, and creates reversible repair plans.
0 stars49 downloads/wk
Reviews
Write oneNobody has reviewed dr-mcp yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
dr-mcp tools (7, 2 write)
write = sends, deletes, buys or postsRead from the package source without running it. The installed server may list more.
apply_patch_planwrite actionApply a patch plan after explicit confirmation. Creates backups before writing.
dr_mcp_cleanupScan and return the highest-value MCP cleanup candidates. Does not apply patches.
dr_mcp_scanOne-step MCP cleanup scan for agent sessions. Checks context weight, stale packages, abandoned repos, duplicates, and safe patch plans.
explain_issueExplain a diagnostic from the current MCP setup scan.
export_reportExport a redacted dr-mcp report as Markdown, HTML, or JSON.
generate_patch_planwrite actionGenerate a reversible patch plan for a scan finding.
scan_mcp_setupScan local MCP configuration files and return a redacted diagnostic report.
Public scan report
scanner v0.1.9 · 2026-09-27 · same rubric, same numbers if you re-run it
- Code scan32 source files scanned25/25
- –Live reliabilityno gateway calls yet and no remote to proben/a
- –Tool poisoningtools not inspected (local package is not executed); not countedn/a
- Auth qualitylocal package, no credentials required12/15
- Maintenancelast push 131 days ago8/15
- Maintainer identityregistry namespace matches repository owner7/10
How dr-mcp compares
Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 80/100
- higher than 60% of 48,255 servers
- Tools
- 7
- the typical server has 7
- Tools that change things
- 29%
- typical: 0%. Fewer write tools means less an agent can break
- Downloads a week
- 49
- more than 41% of 12,191 servers with a package
What its tools let an agent do: 5 read · 1 create · 1 run or deploy
Install dr-mcp in Claude Code, Cursor or VS Code
Runs npx -y @inferensys/dr-mcp on your machine. Read the scan report first; the gateway never runs local packages.
claude mcp add dr-mcp -- npx -y @inferensys/dr-mcp
What the publisher says
From the dr-mcp repository's README, as published. We do not edit it. Read it on GitHub
dr-mcp
Clean your MCPs.
dr-mcp is a local-first MCP audit and cleanup tool for developers who use agentic coding every day. It scans the Model Context Protocol configs your tools load, finds stale servers, abandoned packages, duplicate entries, context-heavy tools, risky permissions, and major upgrades, then generates reversible patch plans.
If you use Claude Code, OpenAI Codex, Cursor, Windsurf, GitHub Copilot, VS Code, Cline, Roo Code, Continue, or Zed across multiple projects, your MCP setup can grow quietly: old demo servers, duplicate GitHub tools, broad filesystem access, package pins behind npm latest, abandoned repositories, and large tool surfaces your agent has to consider before it writes code.
dr-mcp shows what to clean up without deleting anything during scan.
npx @inferensys/dr-mcp cleanupThat starts a local cleanup scan in the current project. You get a redacted report, a ranked list of context-heavy MCPs, package and repository freshness signals, and patch plans you can apply after review.
For the full cleanup scan with package/repo checks and local install-history tracking:
npx @inferensys/dr-mcp cleanup --registry --track-usageInside Claude Code, Codex, Cursor, Windsurf, GitHub Copilot, or any MCP client that exposes server prompts/tools, use drmcpscan or drmcpcleanup. If your client maps MCP prompts into slash commands, use /dr-mcp scan.
Why Clean Your MCPs?
MCP servers are powerful because they put tools directly into an agent's loop. That also means every extra server can add tool-selection noise, context-window pressure, startup failures, stale dependencies, or permissions you no longer intend to grant.
dr-mcp helps with:
- MCP audit: one report across global and project MCP configs.
- MCP cleanup: review candidates for duplicates, dead entries, abandoned packages, long-lived installs, and heavy tool surfaces.
- MCP context hygiene: estimated tool counts and context weight for servers that do not belong in every coding session.
- MCP security review: broad filesystem access, inline secrets, secret-looking args, and broken environment references.
- MCP upgrade planning: semver drift against npm latest, including major upgrades that deserve review.
- MCP repair plans: explicit operations with backups and diffs before writing.
What It Helps You Fix
- Reclaim context. See which MCPs add the biggest tool lists to your agent's prompt.
- Cut unwanted tool calls. Remove old servers your agent keeps considering even when the project does not need them.
- Ditch abandoned MCPs. With --registry, check npm metadata and GitHub activity for archived or stale projects.
- See major upgrades. Find MCP packages pinned far behind npm latest and review upgrade plans.
- Find leftovers. Turn on --track-usage and build a local ledger of MCPs that keep showing up across scans.
- Remove duplicates. Spot the same MCP registered in Claude, Cursor, VS Code, Codex, Windsurf, or project files.
- Fix risky configs. Flag broad filesystem access, inline secrets, secret-looking args, missing commands, dead paths, and broken env refs.
- Stop floating installs. Catch npx packages using latest or no version.
How The Audit Works
dr-mcp uses practical static analysis and package metadata checks:
Shortened. The full README is on GitHub.
Nothing above is checked by us. What we check is on the safety report.
dr-mcp: common questions
- Is dr-mcp server safe?
- Mostly: it is graded B (80/100). Read the dr-mcp safety report
- How do I install dr-mcp?
- It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
- Does dr-mcp need an API key?
- Not as far as the registry entry and our scan can tell: no credentials are declared or required.
- Is dr-mcp maintained?
- The last commit was 168 days ago (2026-05-20). The latest release is v0.3.0.
- How does dr-mcp compare with other MCP servers?
- safety score 80/100 (higher than 60% of 48,255 servers); tools 7 (the typical server has 7); tools that change things 29% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.