HAPI MCP Server
Turn an existing OpenAPI spec into MCP tools so Claude or Cursor can call your REST API without a custom MCP rewrite.
Overview
HAPI MCP is a MCP server for the Build phase that dynamically exposes OpenAPI REST APIs as MCP tools for AI assistants.
What is this MCP server?
- Maps OpenAPI 3.0+ REST operations to MCP tools automatically from spec URL or local file
- Docker quick start: hapimcp/hapi-cli:0.6.0 serve on port 3030 with headless mode
- Env-driven serve: HAPI_OPENAPI and HAPI_URL for custom APIs without editing images
- Reuses existing API auth, rate limits, and governance instead of parallel shadow services
- MIT-licensed HAPI stack with petstore3.swagger.io example spec path in docs
- Example image tag hapimcp/hapi-cli:0.6.0
- Quick-start example binds port 3030
- Supports OpenAPI 3.0+ and OAuth 2.0 Dynamic Client Registration per HAPI docs
What problem does it solve?
Solo builders waste weeks duplicating REST logic in hand-rolled MCP servers every time an agent needs to call their API.
Who is it for?
Indie developers who already have OpenAPI文档 and want Claude Code or Cursor to call production endpoints without a rewrite.
Skip if: Teams with no OpenAPI spec or who only need static, non-HTTP tools without a REST backend.
What do I get? / Deliverables
After you register HAPI MCP with your spec, agents invoke your live REST operations as MCP tools with governance inherited from the API.
- Running MCP server whose tools mirror REST operations from your spec
- Agent-callable endpoints inheriting API auth and rate limits
- Repeatable Docker serve workflow for spec updates
Recommended MCP Servers
Journey fit
Canonical shelf is Build because the server exists to wire production APIs into agent workflows during implementation. Integrations is the right subphase: it bridges OpenAPI contracts to MCP tool surfaces agents can invoke.
How it compares
OpenAPI-to-MCP bridge, not a curated skill pack or single-purpose agent prompt.
Common Questions / FAQ
Who is ai.com.mcp/hapi-mcp for?
It is for solo builders and small teams shipping REST or OpenAPI-backed products who want MCP clients to call those endpoints without maintaining parallel MCP business logic.
When should I use ai.com.mcp/hapi-mcp?
Use it during Build when you are wiring agent tooling to an existing API spec and need tools to update whenever the OpenAPI file changes.
How do I add ai.com.mcp/hapi-mcp to my agent?
Run the published Docker serve command against your spec or HAPI_OPENAPI URL, expose the MCP port, then add that server URL in Claude Code, Cursor, or another MCP client config.