Cyberun Cloud

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.

Cyberun Cloud is the dashboard for running AI work on GPU infrastructure you control. Sign in, create a team, connect one or more agents, and dispatch tasks against workflows. The agents run on your hardware; Cloud schedules them and shows you what’s happening.

​

What you can do here

• Generate images and video from ComfyUI graphs your team publishes, with your team’s agents picking the work up.
• Train 3D Gaussian splat scenes with the Nerfstudio wizard (when enabled on your deployment).
• Run any custom workflow the team has uploaded — text-to-X, enhance, batch, control-net pipelines, anything ComfyUI can do.
• Drive everything from your AI client — Claude Code, Cursor, Windsurf, Codex CLI, VS Code 1.101+ — through the MCP endpoint using an integration key.
• Share results with teammates and partners by inviting them into a team and pointing them at the workflow.

​

Concepts in 60 seconds

• A team holds your agents, workflows, models, credentials, and members. It’s the access boundary.
• A workflow is a reusable template (a ComfyUI graph, a Nerfstudio pipeline, …) plus the agent labels that pick where it can run.
• An agent is a process on a GPU machine your team controls. It dials out to Cyberun and accepts tasks. You bring the hardware.
• A task is one execution of a workflow. Outputs (images, video, models, etc.) appear on the task’s result page.
• Credentials come in three flavours — integration (sk-) for scripts and AI clients, agent (ak-) for GPU hosts, device (dk-) for devices paired through the Cyberun CLI.

​

How it fits together

You sign in through the browser; your AI clients call the same platform through MCP; both end up dispatching tasks to the agents your team runs.

​

Next steps