# task-master-agent

**The Ultimate Task Management & Orchestration Tool for AI Agents**

`task-master-agent` is an npm CLI tool designed to build the ultimate **AI-Native Workflow** for AI coding assistants like Antigravity and Cursor.

It goes beyond simple task management, automating the entire development cycle from requirements analysis to technical design, implementation, and verification.

---

## 🚀 Key Value Propositions

Solves the core problems of **"Context Fragmentation"** and **"Unreliable Verification"** when working with AI agents.

### 1. 🏛️ Architect Mode
Orchestrates the Plan → Do → Check cycle with a lead architect workflow. It independently derives sophisticated tasks from PRDs, creates implementation plans, and performs TDD-style verification.

### 2. ⌨️ AI-Native No-Input CLI
All CLI prompts are built as `selectable lists`. AI agents never need to type `y/n` manually; they can precisely control the tool using arrow keys and Enter, minimizing operational errors.

### 3. 🔍 Smart Context Synchronization
Keeps the agent's context synced so they always know the current state and the next steps. `TASK_BREAKDOWN.md` serves as the single **Source of Truth** for the entire project.

### 4. 🛡️ Strict Auto-Verification & Self-Recovery
Verifies linting, type-checking, building, and **visual fidelity**. When a check fails, it provides specific recovery guides so the AI can analyze and fix the root cause independently.

---

## 📦 Getting Started

```bash
# Run directly to initialize your project without installation.
npx task-master-agent@latest
```

Manage your project settings and workflows through the **Interactive Menu** that appears in your terminal.

---

## 🛠️ AI Agent Workflows (Slash Commands)

Call these commands in your agent's chat window (Antigravity, Cursor, etc.) to delegate work.

| Command | Purpose |
| :--- | :--- |
| `/task-architect` | **[Recommended]** Orchestrates the entire process from PRD analysis to final verification. |
| `/task-next` | Execute the next pending task and perform automated verification. |
| `/task-all` | Sequentially process all remaining tasks in batch. |
| `/write-prd` | Collaborate with the agent to write high-quality PRDs. |
| `/verify-task` | Strictly verify implementation quality and visual fidelity. |

---

## 📈 Best Practices

1.  **Initialize**: Run `task-master-agent init` to set up the project structure.
2.  **Design**: Define requirements with `/write-prd` and generate tasks via `parse-prd`.
3.  **Execute**: Delegate implementation using `/task-architect` or `/task-next`.
4.  **Sync**: Call `sync` during development to keep the agent's context fresh.
5.  **Verify**: Perform final functional and visual checks using the updated `verify-task` workflow.

---

## 📂 Project Structure

- `.agent/`: Repository for the core engine, rules, and automation workflows.
- `.agent/TASK_BREAKDOWN.md`: The central document for tracking task states.
- `.task-master.json`: Project-specific customized configuration.

---

## License

MIT