# Claude Task Master

> AI-driven task management and decomposition for long-running agent projects.

[Canonical HTML page](https://top-mcps.com/mcp/task-master-ai) · [server.json](https://top-mcps.com/mcp/task-master-ai.json) · [methodology](https://top-mcps.com/about/methodology)

## Install

### Claude Desktop — `claude_desktop_config.json`

Paste under mcpServers. Fully quit and reopen Claude after editing.

```json
{
  "mcpServers": {
    "task-master-ai": {
      "command": "npx",
      "args": [
        "-y",
        "task-master-mcp"
      ]
    }
  }
}
```

### Claude Code — `CLI or .mcp.json`

Run from your repo. Commit .mcp.json to share with your team.

```shell
claude mcp add task-master-ai -- npx -y task-master-mcp
```

### Cursor — `.cursor/mcp.json`

Global path: ~/.cursor/mcp.json. Reload window after editing.

```json
{
  "mcpServers": {
    "task-master-ai": {
      "command": "npx",
      "args": [
        "-y",
        "task-master-mcp"
      ]
    }
  }
}
```

### VS Code — `.vscode/mcp.json`

VS Code uses the "servers" key (not "mcpServers").

```jsonc
{
  "servers": {
    "task-master-ai": {
      "command": "npx",
      "args": [
        "-y",
        "task-master-mcp"
      ]
    }
  }
}
```

### Windsurf — `~/.codeium/windsurf/mcp_config.json`

Open via Cascade → hammer icon → Configure.

```json
{
  "mcpServers": {
    "task-master-ai": {
      "command": "npx",
      "args": [
        "-y",
        "task-master-mcp"
      ]
    }
  }
}
```

### Cline — `cline_mcp_settings.json`

Open via the Cline sidebar → MCP Servers → Edit.

```json
{
  "mcpServers": {
    "task-master-ai": {
      "command": "npx",
      "args": [
        "-y",
        "task-master-mcp"
      ]
    }
  }
}
```

### Continue — `~/.continue/config.json`

Continue uses modelContextProtocolServers with a transport block.

```json
{
  "experimental": {
    "modelContextProtocolServers": [
      {
        "transport": {
          "type": "stdio",
          "command": "npx",
          "args": [
            "-y",
            "task-master-mcp"
          ]
        }
      }
    ]
  }
}
```

### Codex CLI — `~/.codex/config.toml`

Codex uses TOML. Each server is a [mcp_servers.<name>] subtable.

```shell
# ~/.codex/config.toml
[mcp_servers.task-master-ai]
command = "npx"
args = [
  "-y",
  "task-master-mcp",
]
```

### Zed — `~/.config/zed/settings.json`

Zed calls them "context_servers". Settings live-reload on save.

```jsonc
{
  "context_servers": {
    "task-master-ai": {
      "command": {
        "path": "npx",
        "args": [
          "-y",
          "task-master-mcp"
        ]
      }
    }
  }
}
```

### ChatGPT — `ChatGPT → Apps directory`

Claude Task Master doesn't ship a hosted HTTPS endpoint today. ChatGPT supports remote MCP servers only — to use this server in ChatGPT you'll need to deploy it to a public HTTPS URL first (e.g. via Cloudflare Workers or Vercel) or wait for an official remote build.

```none

```

## At a glance

- **Maintainer:** Eyal Toledano (community)
- **Transport:** stdio
- **Auth model:** None
- **Required secrets:** None
- **Supported clients:** Claude, Claude Code, Cursor, Any MCP-compatible client
- **License:** MIT
- **Language:** TypeScript
- **Latest version:** latest
- **Last verified:** 2026-05-26
- **GitHub stars:** 27,256 (fetched 2026-05-26T10:50:23.471Z)
- **Score:** 86/100 (rubric 2026-04 — see https://top-mcps.com/about/methodology)
- **Source:** https://github.com/eyaltoledano/claude-task-master

## Security & scope

- **Access scope:** read-write
- **Sandbox:** Reads and writes to a `.taskmaster` directory inside the project root. No network calls.
- **Gotchas:**
  - State files are checked in by default — review before committing sensitive task descriptions.
  - Concurrent edits across multiple agents are not coordinated; one writer at a time.

## Quick answer

**What it does.** Exposes tools to add, expand, complete, and reorder tasks. Stores state on disk so an agent can pick up exactly where it left off across sessions, models, and clients.

**Best for:**
- Multi-week coding projects
- Plan-then-execute agent loops
- Cross-session continuity
- Decomposing vague specs into checklists

**Not for:**
- Single-session tasks
- Team workflows that need collaboration features

## Description

Claude Task Master is an MCP server that turns a high-level project goal into a structured task list, tracks progress, and persists state across sessions. Built specifically for AI coding agents that need to plan, decompose, and execute multi-week initiatives.

## Why it matters

Long-running agent projects fall apart on persistence and decomposition. Task Master is opinionated about both: every task has a status, a parent, and a checklist of subtasks the agent can drive through.

## Key features

- Hierarchical task lists
- State persistence on disk
- Status tracking per task
- Expand vague tasks into subtasks
- Works across Claude/Cursor/VS Code

## FAQ

### How does it differ from a TODO file in the repo?

Task Master is structured — every task has a status, dependencies, and child subtasks the model can iterate. A plain TODO file requires the agent to re-derive structure on every read.

### Where is state stored?

In a `.taskmaster` directory in the project root by default. JSON files, version-controllable, human-readable.

### Does it call an LLM internally?

For decomposition (expanding a task into subtasks), yes — the agent's own model is invoked through the MCP host. No separate API key needed.

## Changelog

- **2026-05-26** — Refreshed install snippets and fact sheet; verified for 2026.
- **2025-02-01** — Initial directory listing.