# SuperCool Docs

> Build with SuperCool. Send your SuperCool agent a message from your own code, terminal or AI app, and get finished work back.

Source: https://supercool.com/docs

SuperCool is one agent that makes finished work from a plain-language request: videos, images, websites, decks, research and writing. These docs cover the ways to reach it from outside the app.

<div class="cards">
<a href="/docs/api"><strong>API</strong>Your agent over HTTPS. One endpoint: send a message, get back the reply, the work it started and the files it made.</a>
<a href="/docs/api/quickstart"><strong>Quickstart</strong>From API key to downloaded video in five minutes, with curl, Python or JavaScript.</a>
<a href="/docs/api/reference"><strong>API reference</strong>Every endpoint, field and error code, generated from the OpenAPI spec.</a>
<a href="/cli"><strong>CLI</strong>Your SuperCool agent in the terminal: <code>supercool ask</code>, attach files, wait for results.</a>
<a href="/mcp"><strong>MCP</strong>Connect SuperCool to Claude, ChatGPT, Cursor and any other MCP client.</a>
<a href="/docs/api/agents"><strong>For AI agents</strong>llms.txt, every page as Markdown, and a prompt snippet for coding agents.</a>
</div>

## One agent everywhere

The API, the CLI and the MCP all talk to the same agent you use in the SuperCool app, by phone and on WhatsApp. It has the same memory and works in the same chats, so work you start from code shows up in the app, and the other way round. Every channel uses the same credits.

## Start here

- New to the API? Read the [introduction](https://supercool.com/docs/api), then follow the [quickstart](https://supercool.com/docs/api/quickstart).
- Building on it? Keep the [API reference](https://supercool.com/docs/api/reference) and [errors](https://supercool.com/docs/api/errors) open.
- Wiring it into an AI agent? Point it at [llms.txt](https://supercool.com/docs/llms.txt) or use the [MCP server](https://supercool.com/mcp).