About

Know what your AI is connected to

AI agents are only as trustworthy as the tools they can call. Squatch MCP makes the Model Context Protocol ecosystem legible: who publishes a server, what it can do, and what changed.

Why we built this

The Model Context Protocol lets an AI assistant read your files, send email, query databases and move money through tools published by anyone. Installing an MCP server is a decision about access, yet most of the information you need to make that decision is scattered, self-reported or invisible. Tool descriptions can change after you connect. Names can imitate well-known publishers. Hidden characters can smuggle instructions into metadata.

Squatch MCP collects that information in one place and shows its source.

Our principles

Evidence, not verdicts. We never call a server "safe", "trusted" or "secure". Each signal says what was checked, the result and the evidence.
Declared is not observed. Publisher statements are labelled "Publisher declared". What we measured is "Squatch observed". Independent proofs are "Squatch verified". Registry facts are "Official MCP Registry".
No numeric trust score. A single number hides more than it reveals. You see the dimensions and decide what matters for your use.
Payment never changes results. Paid plans add capacity and features. They never change verification outcomes, findings, classifications or rankings.
Third-party text is treated as hostile. Names, descriptions and schemas from MCP servers are shown escaped, with hidden characters made visible.
We read, we don't run. The scanner inspects metadata. It never invokes a tool. See Security for exactly what it does.

Where the data comes from

We import public listings from the Official MCP Registry, accept submissions from publishers, and inspect remote MCP endpoints with a sandboxed scanner. Publishers can claim their servers, verify domains and repositories, and respond to findings. Their responses are shown alongside our observations, never in place of them.

Who we are

Squatch MCP is built by SaaSquatch Lab LLC, an Oregon company. We are independent and not affiliated with the Model Context Protocol project or Anthropic. Questions or corrections: hello@squatchmcp.com.