Trust Index

The state of agent tool safety.

An independent index of signed answers for the tools, MCP servers, and packages AI agents connect to — checked from the outside, no repo access, verifiable offline. Not the agent you build; the things it trusts.

19,098tools checked, and how they readplus 33,381 paid x402 endpoints
69.4%Safe to connect27.1%Review before you connect3.5%Do not connect

Where the risk is

Share of checked tools with a critical finding, by kind

Agent skills
6.3%
GitHub repos
5.6%
MCP servers
3.4%
PyPI packages
2.3%
npm packages
1.9%
Rust crates
1.4%

Critical findings over tools checked, per kind. Counted in catalog rows, not unique repos.

From the sandbox

last 30 days

Beyond reading the code, we run packages and MCP servers in an isolated sandbox, call every tool, and plant canary credentials. In the last 30 days: 11,650 runs, 1,117 servers exercised, 203 behavioral findings.

Contacted an undeclared host171
Read-only tool wrote files11
Tool returned a secret in its output10
Tool call crashed the server6
Probed cloud metadata5

Aggregate counts only. Each run is signed as a separate observation shown beside the score. How the sandbox works →

The leaderboards

Ranked from the live catalog — click any tool for its full signed report.

Safest MCP servers

by trust
loading…

Most relied upon

by adoption
loading…

Safest npm packages

by trust
loading…

Safest PyPI packages

by trust
loading…

Check the tool you're about to connect.

Paste any repo, package, or MCP server — get an answer and a signed, offline-verifiable trust score in seconds. Free, no account.