Content
# WordPress Trac MCP Server
A Model Context Protocol (MCP) server that provides AI assistants with comprehensive access to WordPress.org Trac data. Built with TypeScript and deployed on Cloudflare Workers.
**🆕 Now with ChatGPT Deep Research support!**
## Overview
This MCP server transforms WordPress Trac into an AI-accessible knowledge base, enabling intelligent queries about WordPress development, ticket tracking, and code changes. Features dual architecture supporting both standard MCP clients and ChatGPT's Deep Research requirements.
## Features
- **Dual Architecture**: Standard MCP + ChatGPT Deep Research support
- Search 60,000+ WordPress tickets by keywords, components, or status
- Get detailed ticket information including descriptions, status, and metadata
- Access changeset information with full diff content
- Monitor recent WordPress development activity
- Retrieve project metadata like components, milestones, and priorities
- **Intelligent Query Routing**: Automatically detects ticket numbers, revisions, and keywords
- **Smart Caching**: Optimizes fetch operations for better performance
- WordPress-branded UI with official styling
## Available Tools
The server provides two different tool interfaces depending on your client:
### Standard MCP Tools
For Claude Desktop, MCP Inspector, and other standard MCP clients:
#### searchTickets
Search through WordPress Trac tickets with intelligent filtering.
```json
{
"tool": "searchTickets",
"args": {
"query": "REST API performance",
"limit": 10,
"status": "open"
}
}
```
#### getTicket
Retrieve comprehensive information about specific tickets.
```json
{
"tool": "getTicket",
"args": {
"id": 59166,
"includeComments": true
}
}
```
#### getChangeset
Access detailed information about code commits and changes.
```json
{
"tool": "getChangeset",
"args": {
"revision": 55567,
"includeDiff": true,
"diffLimit": 2000
}
}
```
#### getTimeline
Monitor recent WordPress development activity.
```json
{
"tool": "getTimeline",
"args": {
"days": 7,
"limit": 20
}
}
```
#### getTracInfo
Get organizational data like components and milestones.
```json
{
"tool": "getTracInfo",
"args": {
"type": "components"
}
}
```
### ChatGPT Deep Research Tools
For ChatGPT's Deep Research feature (simplified interface):
#### search
Intelligent search that automatically routes to the right data based on your query.
```json
{
"tool": "search",
"args": {
"query": "block editor performance"
}
}
```
**Supported query types:**
- **Keywords**: `"REST API bugs"`, `"media upload issues"`
- **Ticket numbers**: `"#61234"`, `"61234"`
- **Changesets**: `"r58504"`, `"58504"`
- **Recent activity**: `"recent"`, `"timeline"`, `"latest"`
- **Components**: `"Block Editor"`, `"REST API"`
#### fetch
Get detailed information about a specific item by ID.
```json
{
"tool": "fetch",
"args": {
"id": "61234"
}
}
```
**Supported ID formats:**
- Ticket IDs: `"61234"`
- Changeset revisions: `"r58504"`
## Installation
### Deploy to Cloudflare Workers
```bash
# Clone the repository
git clone https://github.com/Jameswlepage/trac-mcp.git
cd trac-mcp
# Install dependencies
npm install
# Login to Cloudflare
wrangler login
# Deploy
npm run deploy
```
### Connect to AI Assistant
#### Standard MCP (Claude Desktop, etc.)
Add to your `claude_desktop_config.json`:
```json
{
"mcpServers": {
"wordpress-trac": {
"command": "npx",
"args": ["mcp-remote", "https://your-worker-url/mcp"]
}
}
}
```
#### ChatGPT Deep Research
ChatGPT uses a different connection method than Claude Desktop. Follow these steps:
1. **Open ChatGPT Settings** → Go to the **Connectors** tab
2. **Add Server** → Import your remote MCP server directly:
```
https://your-worker-url/mcp/chatgpt
```
3. **Enable in Composer** → The server will appear in **Composer** > **Deep Research** tool
4. **Add as Source** → You may need to manually add the server as a research source
For detailed setup instructions, see: [ChatGPT MCP Documentation](https://platform.openai.com/docs/mcp#connect-in-chatgpt)
> **Note**: ChatGPT requires exactly 2 tools (`search` and `fetch`) with simplified schemas. The `/mcp/chatgpt` endpoint is specifically optimized for this requirement.
## Development
### Local Development
```bash
# Start development server
npm run dev
# Test with MCP Inspector
npx @modelcontextprotocol/inspector http://localhost:8787/mcp
```
### Testing
```bash
# Run type checking
npm run type-check
# Run linting
npm run lint
# Test deployment
curl https://your-worker-url/health
```
## Architecture
- **Runtime**: Cloudflare Workers for global edge deployment
- **Language**: TypeScript with Zod validation
- **Protocol**: Model Context Protocol (MCP) for universal AI compatibility
- **APIs**: Public WordPress Trac CSV/RSS endpoints (no authentication required)
## Live Demo
**URL**: https://mcp-server-wporg-trac-staging.a8cai.workers.dev
## License
This project is licensed under the GNU General Public License v2 or later - see the [GPL License](https://www.gnu.org/licenses/gpl-3.0.en.html#license-text) for details.
## Contributing
Contributions are welcome! This server demonstrates how to build production-ready MCP servers with real-world complexity and WordPress integration.
