"Server Fault" matching MCP servers:

Fault injection and reliability scoring for MCP servers. Scans tool output schemas, injects corrupted payloads, and measures contract enforceability. Model-free, deterministic, no LLM required.

Enables AI agents to execute HTTP/REST/GraphQL requests against any endpoint, including private VPN hosts and local services, while logging untruncated headers, request bodies, and raw responses into a local SQLite audit history. Agents can then search that history by keyword, date, method, or status, retrieve full details by ID, and export everything to JSON.