claude-task-master/mcp-server

Task Master MCP Server

This module implements a Model Context Protocol (MCP) server for Task Master, allowing external applications to access Task Master functionality and context through a standardized API.

Features

• MCP-compliant server implementation using FastMCP
• RESTful API for context management
• Authentication and authorization for secure access
• Context storage and retrieval with metadata and tagging
• Context windowing and truncation for handling size limits
• Integration with Task Master for task management operations

Installation

The MCP server is included with Task Master. Install Task Master globally to use the MCP server:

npm install -g task-master-ai

Or use it locally:

npm install task-master-ai

Environment Configuration

The MCP server can be configured using environment variables or a .env file:

Variable	Description	Default
MCP_SERVER_PORT	Port for the MCP server	3000
MCP_SERVER_HOST	Host for the MCP server	localhost
MCP_CONTEXT_DIR	Directory for context storage	./mcp-server/contexts
MCP_API_KEYS_FILE	File for API key storage	./mcp-server/api-keys.json
MCP_JWT_SECRET	Secret for JWT token generation	task-master-mcp-server-secret
MCP_JWT_EXPIRATION	JWT token expiration time	24h
LOG_LEVEL	Logging level (debug, info, warn, error)	info

Getting Started

Starting the Server

Start the MCP server as a standalone process:

npx task-master-mcp-server

Or start it programmatically:

import { TaskMasterMCPServer } from "task-master-ai/mcp-server";

const server = new TaskMasterMCPServer();
await server.start({ port: 3000, host: "localhost" });

Authentication

The MCP server uses API key authentication with JWT tokens for secure access. A default admin API key is generated on first startup and can be found in the api-keys.json file.

To get a JWT token:

curl -X POST http://localhost:3000/auth/token \
  -H "x-api-key: YOUR_API_KEY"

Use the token for subsequent requests:

curl http://localhost:3000/mcp/tools \
  -H "Authorization: Bearer YOUR_JWT_TOKEN"

Creating a New API Key

Admin users can create new API keys:

curl -X POST http://localhost:3000/auth/api-keys \
  -H "Authorization: Bearer ADMIN_JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"clientId": "user1", "role": "user"}'

Available MCP Endpoints

The MCP server implements the following MCP-compliant endpoints:

Context Management

• GET /mcp/context - List all contexts
• POST /mcp/context - Create a new context
• GET /mcp/context/{id} - Get a specific context
• PUT /mcp/context/{id} - Update a context
• DELETE /mcp/context/{id} - Delete a context

Models

• GET /mcp/models - List available models
• GET /mcp/models/{id} - Get model details

Execution

• POST /mcp/execute - Execute an operation with context

Available MCP Tools

The MCP server provides the following tools:

Context Tools

• createContext - Create a new context
• getContext - Retrieve a context by ID
• updateContext - Update an existing context
• deleteContext - Delete a context
• listContexts - List available contexts
• addTags - Add tags to a context
• truncateContext - Truncate a context to a maximum size

Task Master Tools

• listTasks - List tasks from Task Master
• getTaskDetails - Get detailed task information
• executeWithContext - Execute operations using context

Examples

Creating a Context

// Using the MCP client
const client = new MCPClient("http://localhost:3000");
await client.authenticate("YOUR_API_KEY");

const context = await client.createContext("my-context", {
  title: "My Project",
  tasks: ["Implement feature X", "Fix bug Y"],
});

Executing an Operation with Context

// Using the MCP client
const result = await client.execute("generateTask", "my-context", {
  title: "New Task",
  description: "Create a new task based on context",
});

Integration with Other Tools

The Task Master MCP server can be integrated with other MCP-compatible tools and clients:

• LLM applications that support the MCP protocol
• Task management systems that support context-aware operations
• Development environments with MCP integration

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.