TaskWarrior MCP Server: Your Tasks, Meet Your AI

Stop switching between ChatGPT tabs and your terminal to manage tasks. This MCP server connects your AI assistant directly to TaskWarrior, turning task management into natural conversation.

Why This Matters

If you're already using TaskWarrior (or thinking about it), you know the CLI is incredibly powerful but requires constant context switching. You're coding, get an idea, Alt+Tab to terminal, type task add, switch back to your editor. Or you ask Claude to help plan a project, then manually copy tasks into TaskWarrior.

This server eliminates that friction completely. Your AI can query your actual task list, add properly formatted tasks, and mark things complete—all while staying in your conversation flow.

What You Get

Direct TaskWarrior Integration: Your AI assistant talks directly to your local task binary. No sync, no import/export, no duplicate data entry.

Natural Task Management: Instead of memorizing TaskWarrior syntax, just tell Claude "Add a high priority task to call the client about the API changes" and it executes task add priority:H Call client about API changes.

Context-Aware Planning: Ask "What are my current work tasks?" and get your actual TaskWarrior project:work list. Claude can help prioritize based on your real workload, not hypothetical examples.

Smart Task Creation: The server handles TaskWarrior's format requirements—projects, tags, priorities, due dates—so your AI assistant creates properly structured tasks every time.

Real-World Scenarios

Project Planning Sessions: "I need to ship the user auth feature by Friday. Here's what I'm thinking..." Claude sees your existing tasks, suggests additions, and creates them with proper project tags and due dates.

Daily Standup Prep: "What should I focus on today?" pulls your actual pending tasks, filters by priority and project, and helps you plan your day based on real data.

Code Review Follow-ups: "Add tasks for the issues we found in the PR review" creates specific, trackable tasks instead of hoping you'll remember later.

Context Switching Prevention: Working on a feature, get interrupted, tell Claude "Add a task to finish the authentication middleware" without leaving your editor or losing focus.

Technical Implementation

Three core tools handle the essential TaskWarrior operations:

• get_next_tasks: Query pending tasks with project and tag filtering
• add_task: Create tasks with full TaskWarrior metadata (priority, project, tags, due dates)
• mark_task_done: Complete tasks by ID or UUID

The server wraps your existing TaskWarrior setup—same data, same configuration, same workflow. Just with AI assistance.

Getting Started

Install TaskWarrior if you haven't already, then add this to your Claude Desktop config:

{
  "mcpServers": {
    "taskwarrior": {
      "command": "npx",
      "args": ["-y", "mcp-server-taskwarrior"]
    }
  }
}

Your existing TaskWarrior data and configuration work immediately. No migration, no setup complexity.

Why TaskWarrior + AI Works

TaskWarrior's structured approach to task management (projects, tags, priorities, dependencies) gives AI assistants the context they need to help you effectively. Instead of managing generic to-do items, you get intelligent task organization that scales with complex projects.

The combination turns task management from a chore into a collaborative process where your AI helps you think through projects, identify missing pieces, and maintain organization without the cognitive overhead.

This server makes that collaboration seamless—your tasks stay in TaskWarrior where they belong, but your AI can help manage them naturally.

Note: Currently uses TaskWarrior's task IDs which can change when tasks are added/removed. The maintainer plans to switch to stable UUIDs in future versions.