create-mcp-server-kit

## MCP Server Kit: `sortd-mcp-integration`

### Overview
This kit scaffolds a production-ready MCP server (`sortd-mcp-integration`) that connects Sortd's Gmail-based task and workflow management system to AI tools like Claude AI. The server exposes MCP tools for managing Sortd boards, tasks, and workflows directly from AI workflows.

### Key Components
1. **FastAPI Server**: A Python-based MCP server (`main.py`) with endpoints for Sortd API interactions.
2. **Configuration**: `config.yaml` pre-configured for Sortd OAuth2 credentials and API keys.
3. **MCP Tools**: Example tools include:
   - `sortd_get_board_tasks`: Retrieve tasks from a Sortd board by board ID.
   - `sortd_update_task_status`: Update task status (e.g., 'In Progress', 'Done') in Sortd.
   - `sortd_create_followup_email`: Generate and log a follow-up email in Sortd.
4. **Dockerfile**: Containerizes the server for deployment.
5. **README.md**: Step-by-step setup guide with Sortd API authentication steps.

### Example MCP Tool Output
```json
{
  "name": "sortd_update_task_status",
  "description": "Update the status of a task in Sortd.",
  "parameters": {
    "task_id": "task_12345",
    "status": "In Progress",
    "notes": "AI-generated follow-up required"
  }
}
```

### Sample AI Workflow Integration
When integrated with Claude AI, the MCP server enables:
- **Automated Task Management**: AI agents can query Sortd boards to fetch high-priority tasks and update their status based on AI-generated insights.
- **Email Follow-ups**: AI can trigger follow-up emails in Sortd when tasks remain unaddressed for >48 hours.
- **Workflow Automation**: AI can orchestrate multi-step workflows (e.g., lead qualification → task creation → email follow-up).

### Deployment Steps
1. **Clone the Kit**:
   ```bash
   git clone https://github.com/your-repo/sortd-mcp-integration.git
   cd sortd-mcp-integration
   ```
2. **Configure Sortd API**: Add your Sortd OAuth2 credentials to `config.yaml`.
3. **Build & Run**:
   ```bash
   docker build -t sortd-mcp-server .
   docker run -p 8000:8000 sortd-mcp-server
   ```
4. **Connect to AI Tools**: Use the MCP server URI (`http://localhost:8000`) in Claude AI or other MCP-compatible tools.

### Error Handling
- Invalid task IDs return a 404 with a descriptive error.
- Rate limits are handled with exponential backoff.
- Logging is configured for debugging (logs stored in `/var/log/sortd-mcp-server`).

### Next Steps
- Extend the kit with custom MCP tools for Sortd's AI features (e.g., urgency detection).
- Add unit tests for MCP tool validation.
- Deploy to a cloud provider (e.g., AWS ECS) for scalability.
```