Free tools for agent builders

Test what agents can actually discover and use.

Diagnose websites and MCP servers with transparent checks. Every result includes the evidence observed, the fix to make, and the next integration path.

Why these checks are different

Deterministic. The score comes from observable HTTP, HTML, and protocol evidence—not an opaque model opinion.

Safe by default. Targets are restricted to public networks, response sizes are bounded, and the MCP doctor never invokes a tool.

Built to fix. Failed checks explain the exact remediation and connect to the tools and recipes that can implement it.