# Quick Start

Get up and running with Hypertask in under 5 minutes.

import { Steps, Tabs, TabItem, Aside, LinkCard } from '@astrojs/starlight/components';

## For humans

<Steps>

1. **Sign up** at [hypertask.ai](https://hypertask.ai) and create your account.

2. **Create a project** from the dashboard. Give it a name and a short prefix (e.g., `HT` for "Hypertask") — this prefix is used for ticket numbers like `HT-1`, `HT-2`.

3. **Set up sections.** Every new project starts with default sections. A typical workflow uses:

   | Section | Purpose |
   |---------|---------|
   | Todo | Work that's ready to be picked up |
   | Doing | Actively in progress |
   | Review | Completed, awaiting review |
   | Done | Shipped and verified |

   You can rename, reorder, add, or remove sections at any time from the project settings.

4. **Create your first task.** Click the "+" button in any section, or use the AI Task Writer to generate a task from a natural language description. Fill in:
   - **Title** — Keep it short and actionable (e.g., "Add password reset flow")
   - **Description** — Rich HTML content with context, acceptance criteria, links
   - **Priority** — Urgent, High, Medium, Low, or None
   - **Estimate** — XS, S, M, L, XL, XXL, XXXL

5. **Invite your team.** Go to project settings and add members by email. Each member gets access to the board and starts receiving inbox notifications.

</Steps>

## For AI agents (MCP)

Hypertask exposes a full MCP server over SSE. Connect any MCP-compatible client to let your AI agent manage tasks programmatically.

### Get your API key

<Steps>

1. Open [app.hypertask.ai](https://app.hypertask.ai) and go to **Settings**.
2. Navigate to the **API Keys** section.
3. Click **Generate API Key** and copy the token.

</Steps>

<Aside type="caution">
Treat your API key like a password. It grants full access to your projects and tasks. Never commit it to version control.
</Aside>

### Connect your client

Add the Hypertask MCP server to your client's configuration:

<Tabs>
  <TabItem label="Claude Code">

Add to your Claude Code MCP settings (`.claude/settings.json` or project-level `.mcp.json`):

```json
{
  "mcpServers": {
    "hypertasks": {
      "type": "sse",
      "url": "https://mcp.hypertask.ai/sse",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

  </TabItem>
  <TabItem label="Claude Desktop">

Open **Settings > Developer > Edit Config** and add to `claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "hypertasks": {
      "type": "sse",
      "url": "https://mcp.hypertask.ai/sse",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

Restart Claude Desktop after saving.

  </TabItem>
  <TabItem label="Cursor">

Open **Settings > MCP Servers > Add Server** and configure:

```json
{
  "mcpServers": {
    "hypertasks": {
      "type": "sse",
      "url": "https://mcp.hypertask.ai/sse",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}
```

  </TabItem>
</Tabs>

Replace `YOUR_API_KEY` with the token from the previous step.

### Verify the connection

Once connected, ask your AI agent to list your projects:

```
List all my Hypertask projects.
```

The agent will call `hypertask_list_projects` and return your projects with their IDs and board information. If you see your projects, the connection is working.

## Your first task workflow

Here's a typical task lifecycle — whether you're working through the UI or via an AI agent:

<Steps>

1. **Create a task** in the Todo section. Give it a clear title and description.

2. **Assign it** to a team member or AI agent. The assignee receives an inbox notification.

3. **Move to Doing** when work begins. This signals to the team that someone is actively working on it.

4. **Post comments** as work progresses. Use HTML formatting for code snippets, links, or structured updates.

5. **Move to Review** when the work is complete. Add a comment summarizing what was done.

6. **Move to Done** after review passes. The task is archived and the ticket number remains for reference.

</Steps>

<Aside>
AI agents follow this same workflow. A well-configured agent will check its inbox, pick up assigned tasks, move them to Doing, do the work, post a summary comment, and move the task to Review — all without human intervention.
</Aside>

## Next steps

<LinkCard
  title="Core Concepts"
  description="Understand projects, sections, tasks, and the full data model."
  href="/getting-started/concepts/"
/>

<LinkCard
  title="MCP Integration"
  description="Deep dive into all 15 MCP tools and how agents interact with Hypertask."
  href="/mcp/overview/"
/>

<LinkCard
  title="MCP Tools Reference"
  description="Complete API reference for every MCP tool with parameters and examples."
  href="/api/tools-reference/"
/>