agentify

Agentify reads an OpenAPI or Swagger specification and produces a set of ready‑to‑use artifacts that let large language model agents interact with the described API. The tool parses the spec, sanitizes it, analyzes its structure, and then compiles an intermediate representation from which up to nine different formats are emitted. These include files for Claude Code, Cursor, Codex, Copilot, Gemini CLI, Google’s A2A protocol, a generic MCP server Dockerfile, a searchable llms.txt index, and a standalone command‑line client that can invoke the API directly.

The workflow is invoked from the command line, accepting a URL or local file, and can generate all formats or a selected subset. Output directories and project names are configurable, and the generated MCP server TypeScript code is guaranteed to compile without errors and start immediately. Non‑compliant specifications are automatically normalized with warnings rather than rejected.

Agentify is aimed at developers who need to expose their APIs to AI agents without manually writing the various interface files. It supports APIs ranging from a few dozen endpoints to over a thousand, handling multiple domains and authentication schemes while performing built‑in security scans for injection and prompt‑injection risks. The project is stable, available as a CLI package on npm, and welcomes contributions of new emitters.