> ## Documentation Index
> Fetch the complete documentation index at: https://docs.cyberun.cloud/llms.txt
> Use this file to discover all available pages before exploring further.

# Product map

> Which Cyberun surface solves which problem — one-page chooser.

Cyberun has one platform and several ways to use it. Use this
chooser to figure out which to read next.

## Quick chooser

| If your situation is... | Start here |
| - | - |
| "I want to use Cyberun in my browser" | [Cloud](/cloud/overview) |
| "I want to run team workflows on my own GPU" | [Connect an agent](/cloud/guides/connect-agent) |
| "I'm making narrative video with locked characters" | [Studio](/studio/overview) (private alpha) |
| "I want my AI client to drive Cyberun for me" | [MCP](/mcp/overview) |
| "Teach my AI agent the Cyberun layout once" | [AI skill](/skill) |
| "I want to drive Cyberun from the terminal" | [CLI](/cloud/guides/cli) |
| "I'm writing code that calls Cyberun" | [API reference](/api-reference/introduction) |
| "I'm self-hosting / deploying on-prem" | [Platform](/platform/overview) |

## The surfaces, briefly

### Cyberun Cloud — the dashboard

**You want**: a team to run AI workflows on GPU agents, see results
in a browser, and connect external tools through API keys or MCP.

**Read**: [Cloud → Overview](/cloud/overview)

The entry point for most users. Sign in, create a team, connect an
agent, run a task.

### Run workflows on your own GPU

**You want**: to connect your own GPU machine into your team's
pool, making the team's nodes, models, and workflows available
locally, alongside whatever ComfyUI ecosystem you already have.

**Read**: [Connect an agent](/cloud/guides/connect-agent), or the
[Cyberun CLI](/cloud/guides/cli) if you live in a terminal.

The agent pairs to Cloud and your GPU joins the team's pool; tasks
you didn't submit may land on your machine (with your consent), and
tasks you did submit can prefer your machine.

### Cyberun Studio — the creator workspace

**You want**: to make narrative video where the same character,
location, and props stay consistent across many shots, with a
project model built around stages, beats, and takes rather than
isolated clips.

**Read**: [Studio → Overview](/studio/overview)

Studio is in **private alpha**. The first workspace, VideoMaker,
covers the full pipeline end-to-end: script → world and assets →
shots and takes → music, voice, sound effects → exported cut. The
Story → World → Assets → Shots → Dubbing structure mirrors how a
small film crew divides labour. Shares sign-in and the team model
with Cyberun Cloud.

### MCP & Skill — driving Cyberun from AI clients

**You want**: your AI client (Claude Code, Cursor, Windsurf,
Codex CLI, VS Code 1.101+) to list workflows, submit tasks, and
fetch results — without you wiring HTTP calls by hand.

**Read**: [MCP → Overview](/mcp/overview)

Cyberun ships an MCP server in the cloud that speaks Streamable
HTTP (the current MCP standard). The **Cyberun skill** is a
separate, lightweight thing: a markdown file your agent loads once
so it knows the platform's conventions.

### Cyberun CLI — the terminal

**You want**: to drive your team's workflows from a shell — submit a
task, poll it, download the result — or wire Cyberun into CI, without
leaving the terminal.

**Read**: [Cloud → CLI](/cloud/guides/cli)

The `cyberun` binary signs in with `cyberun auth login`, runs
workflows with `cyberun run`, and can turn the machine into an agent
with `cyberun agent serve`. It ships on its own release track —
install it from the [download page](/install).

### Platform — self-host, federation, deployment

**You want**: to run Cyberun inside your own datacenter, your
VPC, an air-gapped lab, or a sovereign cloud. Or you're
integrating against the platform from your own code and need to
understand identity, license, and the gateway protocol.

**Read**: [Platform → Overview](/platform/overview) and the
[API reference](/api-reference/introduction).

Aimed at operators and partners — multi-site Kubernetes
federation, OIDC identity, license administration, agent runtime
protocol.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.