Outworx Docs launches hosted API docs with per-project MCP servers
Outworx Docs is a hosted API documentation platform that generates an MCP server per project, enabling Claude, Cursor, Cline, and other MCP-aware tools to interact directly with API specs.
Score breakdown
Teams publishing API docs get an MCP server automatically, meaning AI coding assistants like Cursor and Claude can query live specs, generate typed clients, and run real API calls without manual copy-pasting of documentation.
- 01Each Outworx Docs project ships with a dedicated MCP server accessible via one URL and one token.
- 02The MCP server exposes 12 tools: endpoint listing/search, full schema retrieval, TypeScript type generation, live API calls, and spec-grounded Q&A.
- 03Self-healing types infer response shapes from real traffic when the spec is incomplete, and drift detection invalidates cached types when the API changes.
Outworx Docs is a hosted API documentation platform that accepts OpenAPI 3.x, Swagger 2.0, or GraphQL SDL specs — via file upload, pasted content, or a URL that is re-fetched on a schedule — and produces a live docs site with no config files or build steps. The post by aemadeldin describes the platform as going from upload to live docs in under a minute, with features including custom branding, custom domains with auto-provisioned SSL, dark and light mode, password protection, multi-version API support, an analytics dashboard, and SEO optimization.
Each project gets one URL and one token that plugs into Claude Desktop, Cursor, Cline, or any MCP-aware tool.
The headline feature is a per-project MCP server. Each project gets one URL and one token that plugs into Claude Desktop, Cursor, Cline, or any MCP-aware tool. The MCP server exposes 12 tools: listing and searching endpoints, retrieving full schemas, generating TypeScript types for the whole spec or a single endpoint, running live calls with user-supplied credentials, and spec-grounded Q&A. The platform also includes a "self-healing types" mechanism — when a spec under-documents a response, MCP calls infer the shape from real traffic, merging samples into a stored schema. Drift detection is built in: if the API returns a new field or changes a type, the next call surfaces it, widens the union, and invalidates cached types so they regenerate.
The AI chat assistant appears as a floating "Ask AI" button on every docs site and is trained exclusively on the uploaded spec, with the platform claiming it admits when something is not documented rather than hallucinating. Every endpoint reference in a chat answer is a clickable pill that deep-links to that endpoint in the docs. All chats are logged in a dashboard so API owners can see what visitors ask and identify endpoints needing better descriptions. The interactive Try It playground supports real requests, including to localhost via a lightweight proxy, with shared auth headers persisting across endpoints. Code examples are provided in cURL, Fetch, Axios, and Python, with AI-generated examples available in additional languages and frameworks.
Key facts
- 01Each Outworx Docs project ships with a dedicated MCP server accessible via one URL and one token.
- 02The MCP server exposes 12 tools: endpoint listing/search, full schema retrieval, TypeScript type generation, live API calls, and spec-grounded Q&A.
- 03Self-healing types infer response shapes from real traffic when the spec is incomplete, and drift detection invalidates cached types when the API changes.
- 04