GCC Docs MCP Server

An MCP (Model Context Protocol) server that indexes Singapore Government GCC documentation and exposes search tools for AI assistants. Works with Kiro, Claude Desktop, Cline, and any MCP-compatible client.

Quick Start

Prerequisites

• Python 3.11+
• Git
• gcc-docs-sourceoftruth cloned as a sibling folder

Setup

# Clone both repos side by side
git clone https://github.com/Fredwong76/gcc-docs-sourceoftruth.git sourceoftruth
git clone https://github.com/Fredwong76/gcc-docs-mcp-server.git gcc-docs-mcp-server

# Install the MCP server
cd gcc-docs-mcp-server
pip install -e .

Configure your MCP client

Add to your MCP settings (e.g. ~/.kiro/settings/mcp.json):

{
  "mcpServers": {
    "gcc-docs": {
      "command": "python",
      "args": ["-m", "server.cli"],
      "cwd": "/absolute/path/to/gcc-docs-mcp-server",
      "disabled": false,
      "autoApprove": ["search_docs", "get_page", "list_topics"]
    }
  }
}

Replace /absolute/path/to/gcc-docs-mcp-server with the actual path on your machine.

What it does

On startup the server:

1. Runs git pull on the sourceoftruth folder (gets latest docs from collaborators)
2. Discovers all markdown files in the sourceoftruth folder
3. Builds a full-text search index (Whoosh/BM25F)
4. Exposes 3 MCP tools via stdio transport

Tools

Tool	Description	Example
search_docs	Full-text search across all indexed docs	"security groups port 22"
get_page	Retrieve full content of a specific page	"TechPass/reset-password.md"
list_topics	Browse documentation hierarchy	parent="TechPass"

Configuration

The server finds the sourceoftruth folder using this precedence:

1. --crawl-store <path> CLI argument
2. SOURCEOFTRUTH_PATH environment variable
3. Default: ../sourceoftruth (sibling directory)

Folder Structure

parent-folder/
  sourceoftruth/         <-- git clone of gcc-docs-sourceoftruth
    GCC/
    TechPass/
    Cloudscape/
    ...
  gcc-docs-mcp-server/   <-- this repo
    server/              <-- MCP server code
    crawler/             <-- Playwright-based crawler (optional)
    postprocessing/      <-- PDF-to-MD converter

Development

# Install with dev dependencies
pip install -e ".[dev]"

# Run tests
pytest tests/ --ignore=tests/integration

# Run the server directly
python -m server.cli

Adding Documents

See the sourceoftruth README for how to contribute documentation.