> ## Documentation Index
> Fetch the complete documentation index at: https://dec5a79-spikonado.manicule.dev/docs/llms.txt
> Use this file to discover all available pages before exploring further.
> For every page in one file, fetch https://dec5a79-spikonado.manicule.dev/docs/llms-full.txt (large).
> To search and read these docs as tools, connect the MCP server at https://dec5a79-spikonado.manicule.dev/docs/api/mcp.
> Every agent surface: https://dec5a79-spikonado.manicule.dev/docs/.well-known/llms-manifest.json

# Quickstart (/docs/quickstart)

<!-- agent-signals: reading_time_min: 1 · est_tokens: 437 · updated: 2026-11-12 -->
Related: [Introduction](/docs/index.md), [Install](/docs/install.md), [Troubleshooting](/docs/troubleshooting.md)

> Install Sprocket, sign in, open a project folder, and send the agent its first task.

<Steps>
  <Step title="Install Sprocket">
    Download the desktop app for your OS from the [latest release](https://github.com/spikonado/sprocket/releases/latest) and open it.

    To skip the install, run Sprocket in your browser with [Node.js](https://nodejs.org/) 20.11 or newer:

    ```sh
    npx usesprocket
    ```

    See [Install](/docs/install) for every option.
  </Step>

<Step title="Sign in">Sign in with your Sprocket account. New accounts start on the Free plan, which includes some AI usage each month. You don't need an API key.</Step>

<Step title="Open a project">Choose **Add project** and pick a folder. Sprocket groups threads by project, so every clone with the same git `origin` remote shares one thread list.</Step>

  <Step title="Send a task">
    Describe what you want in the composer and press Enter. For example:

    ```text
    Find out why the tests in this repo fail and fix them.
    ```

    The agent reads files, runs commands, and replies in the thread. Press **Stop generation** to end a run early.
  </Step>
</Steps>

## Run a task from your terminal

Install the CLI, sign in, and run a task in your project folder:

```sh
npm install -g usesprocket
sprocket login
sprocket run "Fix the failing tests"
```

`sprocket run` prints the agent's final answer to stdout. See the [CLI reference](/docs/reference/cli).

## Next steps

* [Add AGENTS.md instructions and skills](/docs/use/instructions-and-skills) so the agent follows your project's rules.
* [Connect your own models](/docs/use/models-and-providers) with API keys or a ChatGPT or Claude subscription.
* [Install the KiCad connector](/docs/use/hardware) for schematic and PCB work.