task-master-install

# Task Master Install

## Overview

Task-master is an AI-powered task management system for breaking down complex projects into
manageable tasks and subtasks. It supports PRD parsing, complexity analysis, and
specification-driven development workflows.

## Installation

### Step 1: Check if already installed

```bash
task-master --version
```

If task-master is already installed and the project has a `.taskmaster/` folder, you are finished.

### Step 2: If you need to install Task-Master

**Global installation (recommended):**

```bash
npm install -g task-master-ai
```

**Local/npx alternative:**

```bash
npx task-master-ai init
```

### Step 3: If you don't have a `.taskmaster/` folder in your project

```bash
task-master init
```

**Common flags:**

- `--name <name>` - Set project name
- `--description <text>` - Set project description
- `--version <version>` - Set initial version (e.g., '0.1.0')
- `--rules <profiles>` - Specify rule profiles (e.g., `cursor,windsurf`)
- `-y, --yes` - Skip prompts, use defaults

**Example:**

```bash
task-master init --name "My Project" --description "AI-powered web app" --version "0.1.0" --rules cursor,windsurf --yes
```

## What Happens During Init

Task-master init automatically creates:

- `.taskmaster/` directory structure (config, tasks, docs, reports, templates)
- Rule files for AI coding assistants (`.cursor/rules/`, etc.)
- Configures AI provider and models to use: Cortex-Code, Claude-Code, Gemini-CLI, Codex-CLI, API,
  etc.

**The bootstrapped rules guide all future task-master workflows** - no additional setup needed.

## Installing MCP Server for AI Agents

To enable AI agents to interact with task-master through MCP tools:

**Option 1: Using Cortex (Recommended)**

```bash
# Using npx (no global install required)
cortex mcp add --args '-y, task-master-ai, mcp' taskmaster npx
snova mcp add --args "run, -i, --rm, -e, DOCUMENT_URL, -e, DOCUMENT_TOKEN, -e, DOCUMENT_ID, -e, RUNTIME_URL, -e, RUNTIME_TOKEN, -e, ALLOW_IMG_OUTPUT, datalayer/jupyter-mcp-server:latest" --env "DOCUMENT_URL=http://host.docker.internal:8888, DOCUMENT_TOKEN=jupyter_mcp_token, DOCUMENT_ID=notebooks/model_training.ipynb, RUNTIME_URL=http://host.docker.internal:8888, RUNTIME_TOKEN=jupyter_mcp_token, ALLOW_IMG_OUTPUT=true" jupyter-mcp-server docker

# Or if task-master is installed globally
cortex mcp add taskmaster task-master --args "mcp"
```

**Option 2: Manual Configuration**

Add to your MCP configuration (e.g., `.cursor/mcp.json` or `claude_desktop_config.json`):

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

**Or if task-master is installed globally:**

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

After configuration, restart your AI agent to connect to the task-master MCP server.

## Troubleshooting

**Node.js not found** - Install Node.js v16+ using:

- Windows (with admin): `winget install OpenJS.NodeJS`
- macOS: `brew install node`
- Linux: `sudo apt install nodejs npm` (Debian/Ubuntu) or `sudo yum install nodejs` (RHEL/CentOS)
- Download installer: [nodejs.org](https://nodejs.org/)

**Installation issues** - Uninstall and reinstall globally:

```bash
npm uninstall -g task-master-ai
npm install -g task-master-ai
```

Then restart terminal and verify with `task-master --version`

**Permission errors:**

- Unix/macOS: Install via nvm using
  `curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash`, restart
  terminal, and run `nvm install 24`
- Windows without admin: Use Node.js installer's "Install for me only" option