Keymaster Mcp

Keymaster Mcp is an open-source trust infrastructure for human–AI teams, focused on credential security and external outcome verification. It allows AI agents to discover and validate approved capabilities (such as credential health) without exposing actual credentials to the AI or model, and enforces acceptance of results only when supported by independent, observable evidence. Designed to prevent credential leaks and enforce bounded trust, Keymaster integrates with the Model Context Protocol (MCP) and is best suited for secure, auditable agent workflows where non-disclosure and evidence-driven automation are required.