Design partners · weekly feedback, three months
Talk to usStart free

Free tool · no signup

MCP Scanner: what can this server’s tools do?

Paste the URL of a public MCP server. We connect, list its tools and flag the ones that may delete, write, export, send data out, run code, move money or touch credentials. We only run initialize and tools/list: no tool is ever called.

By scanning, you confirm you own this server or are authorized to access it. The scanner reads only what the server publishes (initialize and tools/list), never calls a tool, and stores neither the URL nor the result. Privacy

How it works

  • We send initialize and tools/list to the URL (MCP Streamable HTTP). Tools are never called.
  • Each tool is classified from its name, description, input schema and annotations. The score is 0-100: the riskiest tool sets it, and each other flagged tool adds a little.
  • A GitHub repo URL works when the repo has a server.json (MCP registry format) listing a remote streamable-http URL.
  • https only. Private, local and cloud-metadata addresses are refused, also after DNS and on redirects. Short timeouts and a 1 MB size cap.

Heuristic, not a verdict. Results are based only on what the server says about its tools, which can be incomplete or wrong. A low score does not mean a server is safe, and a high score is not a vulnerability finding. This is not a security audit or certification, and results are provided as is, without warranty. Vulnify does not scan code for vulnerabilities.

Acceptable use. Only scan servers you own or are authorized to access. Do not use the scanner to probe, overload or gain access to systems, or to evade rate limits. We may block abusive use. Requests come from Vulnify’s servers with the User-Agent VulnifyMcpScanner/1.0.

Server operators: questions or abuse reports go to security@vulnify.io. To exclude your server from the scanner, write from an address at its domain; we process requests within 5 business days.