Back to MCP Directory
Taskwarrior
Model Context Protocol moderate risk

Taskwarrior

MCP server for TaskWarrior, enabling task management (view, add, complete) via natural language, but relies on local TaskWarrior install and unstable task IDs.

Connections & Capabilities

Connects To

GitHubTwitter

Capabilities

readwriteexec

Quickstart

Install

npm install -g mcp-server-taskwarrior

Config

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

Exposed MCP Tools (3)

safe
get_next_tasks

Retrieves a list of pending tasks, optionally filtered by project and tags.

Read-only operation, no modification of data.

moderate
add_task

Adds a new task to TaskWarrior with a description, due date, priority, project, and tags.

Adds new data, but does not modify existing tasks.

moderate
mark_task_done

Marks a task as completed using its ID or UUID.

Modifies the state of a task, but is generally non-destructive.

Safety Assessment

This server provides a convenient interface to TaskWarrior, but its reliance on local execution and unstable IDs introduces moderate risks. It's relatively safe for personal use, but caution is advised when integrating it into more complex or shared environments.

  • Requires local TaskWarrior installation, limiting external exposure.
  • No external network access beyond the MCP client.
  • Clear tool definitions with specified parameters.
  • Limited set of operations: view, add, and mark tasks as complete.
  • Relies on potentially unstable task IDs for task identification.
  • Executes local `task` binary, inheriting any vulnerabilities in TaskWarrior.
  • No built-in authentication or authorization mechanisms.
  • Lack of input validation could lead to command injection if not carefully handled.
Taskwarrior Review, Pricing & Alternatives (2026) | Flaex AI