Home / MCP / Task Master MCP Server

Task Master MCP Server

Provides an MCP server that runs Task Master to manage AI-driven development tasks via editors and CLIs.

typescript

24.1kstars

View on GitHub

Installation

Add the following to your MCP client configuration file.

Configuration

View docs

{
    "mcpServers": {
        "task_master_ai": {
            "command": "npx",
            "args": [
                "-y",
                "task-master-ai"
            ],
            "env": {
                "ANTHROPIC_API_KEY": "YOUR_ANTHROPIC_API_KEY_HERE",
                "PERPLEXITY_API_KEY": "YOUR_PERPLEXITY_API_KEY_HERE",
                "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY_HERE",
                "GOOGLE_API_KEY": "YOUR_GOOGLE_API_KEY_HERE",
                "MISTRAL_API_KEY": "YOUR_MISTRAL_API_KEY_HERE",
                "GROQ_API_KEY": "YOUR_GROQ_KEY_HERE",
                "OPENROUTER_API_KEY": "YOUR_OPENROUTER_KEY_HERE",
                "XAI_API_KEY": "YOUR_XAI_KEY_HERE",
                "AZURE_OPENAI_API_KEY": "YOUR_AZURE_OPENAI_KEY_HERE",
                "OLLAMA_API_KEY": "YOUR_OLLAMA_API_KEY_HERE"
            }
        }
    }
}

You run Task Master as an MCP server to manage AI-driven development with editors and IDEs. This server lets you offload task planning, parsing PRDs, and task expansion to an AI-powered workflow, while keeping control in your editor or CLI. It supports runtime configuration of multiple AI providers and can scale down token usage by selecting tool loading modes.

How to use

Use an MCP client in your editor or CLI to connect to the Task Master server. You will configure the MCP connection, start the local server, and then interact with Task Master through your editor’s AI chat pane or CLI prompts. You can set which AI models to use, initialize a project, and guide the AI with a detailed Product Requirements Document (PRD) to generate and manage tasks. You can also adjust which tools the server loads to optimize performance.

Key usage patterns include setting up the MCP connection in your editor, starting Task Master, defining your models, and then querying the AI for task parsing, planning, and expansion. You can instruct the AI to parse a PRD, plan the next task, or research new information with project context. You can also move tasks between tags, generate new tasks, and produce complexity reports as your project evolves.

How to install

Prerequisites: ensure you have Node.js and npm installed on your system. You may also install the MCP client in your editor or use the command line interface to run Task Master.

Step-by-step commands to set up and run Task Master locally via the CLI:

Additional configuration and notes

Configure your environment with API keys for the AI providers you wish to use. You can enable different models for main, research, and fallback roles. You can also customize which tools load to optimize token usage.

Tool loading can be tuned to balance feature availability with context usage. The default mode loads all tools, but you can switch to standard, core (lean), or a custom set to fit your project needs.

Troubleshooting: if the MCP server does not respond, verify your API keys are correctly configured and restart the editor or CLI session.

Security note: protect your API keys and limit access to your development environment. Use environment variable management to keep keys out of version control.

Troubleshooting

If you encounter initialization issues, run the initialization script directly with Node to isolate problems. You can also clone the project and run the initialization script locally to ensure the setup works in your environment.

Available tools

get_tasks

Retrieve the list of current tasks with status and metadata

next_task

Identify and return the next task to work on based on project context and dependencies

get_task

Fetch a specific task by ID to view details and subtasks

set_task_status

Update the status of a task or subtask and propagate changes as needed

update_subtask

Modify details within a subtask, including description, due date, and dependencies

parse_prd

Parse a PRD file to generate structured tasks and milestones

expand_task

Expand a task into smaller subtasks with recommended steps

initialize_project

Set up project structure, templates, and initial configurations

analyze_project_complexity

Assess project complexity to guide planning and resourcing

expand_all

Expand all tasks in the backlog into actionable subtasks

add_subtask

Add a new subtask under a parent task with dependencies and tags

remove_task

Remove a task from the plan with dependency checks

generate

Generate new tasks or PRD-derived items based on input prompts

add_task

Add a new top-level task with metadata and context

complexity_report

Produce a report detailing task complexity and effort estimates