I built OpenTrust, an open-source trust and identity layer for AI agent tools. OpenTrust is designed to give AI agents a standardized way to understand whether a tool is safe to use before calling it. The platform uses “Agent Tool Passports” to declare a tool’s identity, permissions, trust level, version history, source formats, and review status across ecosystems like MCP servers, OpenAI functions, LangChain tools, OpenAPI endpoints, npm packages, and PyPI packages. I built the full system myself, including the passport schema, registry API, web frontend, CLI, SDKs, badge generator, manifest validator, payment contract interfaces, and the Hands Body and Feet MCP capability server. The system includes trust-level enforcement, permission scoping, signed passports, revocation support, spend caps, kill switch controls, GitHub identity verification, wallet verification, and machine-readable payment metadata. OpenTrust was built around a simple belief: as AI agents gain the ability to access files, wallets, APIs, terminals, infrastructure, and real-world services, they need a trust layer that is open, verifiable, and framework-agnostic.