Docy MCP Server: Real-Time Technical Documentation Access

Overview

Docy MCP Server is a lightweight server that implements the Model Context Protocol (MCP) to provide AI models with reliable, real-time access to technical documentation. By indexing and serving documentation from local files, Git repositories, and HTTP endpoints, it removes dependence on stale crawled data and external rate-limited APIs. This makes it practical to power assistant workflows that need up-to-date, link-safe answers about code, APIs, and internal docs.

Designed for developers and platform engineers, the server exposes a simple HTTP interface and optional CLI tools to ingest, index, and serve documentation. It fits into CI, chatbots, IDE integrations, or any LLM-based assistant that must reference authoritative docs during inference.

Features

• Implements Model Context Protocol (MCP) for serving structured context to models
• Live document ingestion from:
  • Local file systems
  • Git repositories (GitHub/GitLab)
  • Remote HTTP/markdown endpoints
• Link resolution and sanitization to avoid broken references
• Optional vector/text indexing for fast, semantic retrieval
• Docker and CLI-first deployment, configurable with environment variables or YAML
• Simple HTTP API to query context, list sources, and check health
• Self-hosted to avoid third-party rate limits and to keep internal docs private

Installation / Configuration

Clone the repository and run with Docker or from source.

Clone the repo:

git clone https://github.com/oborchers/mcp-server-docy.git
cd mcp-server-docy

Run with Docker:

# Build and run locally
docker build -t docy-mcp .
docker run -p 8080:8080 \
  -e MCP_PORT=8080 \
  -e DOCY_INDEX_PATH=/data/index \
  -v $(pwd)/docs:/data/docs \
  docy-mcp

Run with docker-compose:

version: "3.8"
services:
  docy:
    image: oborchers/mcp-server-docy:latest
    ports:
      - "8080:8080"
    environment:
      - MCP_PORT=8080
      - DOCY_INDEX_PATH=/data/index
    volumes:
      - ./docs:/data/docs

docker-compose up -d

Example environment variables (table):

Example YAML source configuration (sources.yaml):

sources:
  - type: git
    url: https://github.com/your-org/your-repo.git
    branch: main
    path: docs/
  - type: filesystem
    path: /var/docs/internal
  - type: http
    url: https://example.com/openapi.yaml

After configuring sources, trigger ingestion:

# If the project includes a CLI
./bin/docy ingest --config sources.yaml

Available Resources

Typical HTTP endpoints (adjust to your deployment):

Example MCP-style request (curl):

curl -X POST http://localhost:8080/mcp/context \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4.1",
    "query": "How do I configure OAuth for the API client?",
    "max_context_items": 5
  }'

Example response (abridged):

{
  "context": [
    { "title": "OAuth Setup", "snippet": "To configure OAuth ...", "source": "https://..." },
    ...
  ],
  "metadata": { "retrieval_time_ms": 45 }
}

Use Cases

• IDE assistant: An extension calls /mcp/context with the current file and cursor context to fetch the most recent API docs and examples, enabling precise autocomplete and in-editor help.
• ChatOps and support bots: A chat assistant uses the server to retrieve definitive docs and code snippets when answering questions from developers, avoiding stale references.
• CI documentation checks: Integrate the server into CI to validate that documentation links in PRs resolve and that code examples match the latest API surface.
• Private knowledge base for LLMs: Host internal RFCs, runbooks, and API guides behind your VPC and let LLMs query them without exposing content externally or hitting rate limits.

Getting Help & Contributing

• Repository: https://github.com/oborchers/mcp-server-docy
• Common contribution paths:
  • Add new source adapters (e.g., Confluence, S3)
  • Improve indexing (incremental updates, semantic embeddings)
  • Add authentication/ACLs for production deployments

If you need a quick start, run the Docker image locally and point DOCY_SOURCES to a