Documentation Context Provider MCP Server

This MCP server crawls documentation starting from a given URL, summarizes the combined text content using the Gemini API, and makes it available as context.

Features

• Crawls the starting URL and up to 10 linked pages within the same documentation section.
• Uses Puppeteer for robust web page fetching.
• Summarizes the combined text content using the Google Gemini API (gemini-2.5-flash-preview-0417).
• Optionally reads a package.json file to provide dependency context (though this context isn't currently used in the summarization prompt).
• Exposes a get_doc_summary tool for use with MCP clients like Cline or Claude Desktop.

Prerequisites

• Node.js (version 18 or later recommended)
• npm (usually included with Node.js)
• Git

Installation

1. Clone the repository:

   git clone https://github.com/godzilla675/doc-context-provider-mcp.git # Replace with your actual repo URL if different
   cd doc-context-provider-mcp # Or your chosen directory name

2. Install dependencies:

   npm install

3. Build the server:

   npm run build

   This compiles the TypeScript code into JavaScript in the build/ directory.

Configuration

1. Get a Gemini API Key:

   • Go to Google AI Studio.
   • Sign in and create an API key.
   • Important: Keep this key secure and do not share it publicly.

2. Configure your MCP Client: You need to tell your MCP client (like Cline in VS Code or the Claude Desktop app) how to run this server. Find the appropriate settings file:

   • Cline (VS Code): C:\Users\Ahmed\AppData\Roaming\Code\User\globalStorage\saoudrizwan.claude-dev\settings\cline_mcp_settings.json
   • Claude Desktop (Windows): Typically C:\Users\YourUsername\AppData\Roaming\Claude\claude_desktop_config.json (Replace YourUsername)
   • Claude Desktop (macOS): Typically ~/Library/Application Support/Claude/claude_desktop_config.json

   Open the JSON file and add the following configuration block inside the main "mcpServers": { ... } object. Make sure to replace "YOUR_GEMINI_API_KEY" with the key you obtained, and adjust the path in "args" if you cloned the repository to a different location.

   "doc-context-provider": {
     "command": "node",
     "args": [
       // IMPORTANT: Update this path if you cloned the repo elsewhere!
       "C:\\Users\\Ahmed\\Documents\\Cline\\MCP\\doc-context-provider\\build\\index.js"
     ],
     "env": {
       // IMPORTANT: Replace with your actual API key
       "GEMINI_API_KEY": "YOUR_GEMINI_API_KEY"
     },
     "disabled": false,
     "autoApprove": [
         "get_doc_summary" // Optional: Auto-approve the tool if desired
     ],
     "timeout": 120, // Increased timeout due to crawling/puppeteer
     "transportType": "stdio"
   }

   Ensure you add a comma (,) before this block if it's not the last entry in mcpServers.

3. Restart your MCP Client: Restart VS Code (if using Cline) or the Claude Desktop app for the changes to take effect. The server should now connect automatically.

Usage

Once configured and connected, you can ask your AI assistant (like Cline) to use the tool:

"Get a summary of the documentation at https://react.dev/learn"

Or, if you have a relevant package.json on your Desktop:

"Summarize https://nodejs.org/api/fs.html using package.json"

Limitations

• Crawl Depth: Only crawls links found directly on the starting page (up to 10 relevant links). It does not perform a deep, recursive crawl of the entire site.
• Code Examples: This version focuses on summarizing text and does not extract separate code examples due to the complexity of tracking them across multiple crawled pages.
• Speed: Crawling multiple pages with Puppeteer can be slower than fetching a single page. Summarization time depends on the amount of text processed.
• Dynamic Content: May struggle with sites heavily reliant on client-side JavaScript rendering if content isn't present in the initial HTML or rendered quickly.