ax-eval

Test whether AI agents can use your product — drop in an OpenAPI or GraphQL spec to generate a reviewed L1–L4 task ladder, run it across API/CLI/SDK/MCP surfaces at standard medium effort, and verify with programmatic outcome verification.