Search agents, prompts, tools and users
Jose Cañete
Design a defensive test suite for prompt injection and agent reliability. Covers instruction hierarchy, untrusted conten...
Turn an agent specification, prompt, tools and acceptance criteria into a practical evaluation suite. Produces a coverag...
Review tool and function definitions before they reach production. Detects ambiguous descriptions, invalid or weak JSON ...
Audit an MCP server specification before integration or release. Finds contract gaps, ambiguous tool definitions, unsafe...
Audit any pull request before merge. PR/CI Preflight Auditor analyzes code changes, diffs, tests and CI configuration to...