Feature Request: Integrate MCP Server #1183
Description
Feature Request
Is your feature request related to a problem or unsupported use case? Please describe.
Many product and engineering teams want to generate, query, and update content within docs using programmable interfaces or AI workflows. Currently, there’s no built-in mechanism to support dynamic content creation (e.g., generating specs from tasks), smart search, automated task management, or reporting pipelines — all of which are critical in modern documentation ecosystems.
Describe the solution you’d like
Introduce a lightweight MCP (Multipurpose Content Processor) server mode in docs. This would expose a set of HTTP endpoints to power structured operations like:
- POST /mcp/doc → Generate documentation (e.g., PRDs, tech specs) from project data
- POST /mcp/search → Perform AI-assisted search across the documentation corpus
- POST /mcp/report → Generate summaries, release notes, or weekly digests
This API-driven layer would allow teams to automate and integrate advanced workflows using their own AI models, agents, or CLI tools.
While this concept draws light inspiration from Notion’s MCP spec, the goal here is flexibility — not dependency — and could follow a simpler, pluggable design.
Describe alternatives you’ve considered
- Using external automation scripts that manipulate local files or commit changes directly — functional but inconsistent and hard to scale.
- Integrating with third-party MCP-like services — possible, but reduces control and increases platform reliance.
- Embedding GPT/AI tools manually into docs pipelines — fragmented experience and no standard API layer.
Discovery, Documentation, Adoption, Migration Strategy
- Enable via config flag (enable_mcp: true) in docs.yaml or equivalent.
- Provide an OpenAPI schema for the MCP server endpoints to support SDK generation.
- Document request/response formats for each endpoint with example payloads.
- Allow the MCP module to be extended or overridden with custom handlers.
- Adoption strategy could include CLI tooling or GitHub Actions to call MCP endpoints as part of CI workflows.