Reference
Frequently asked questions
Quick answers covering setup, agents, security, billing, and recovery.
Find an answer
Every feature guide ends with its own detailed FAQ. Use documentation search for a command, setting, error, or workflow, then follow the result directly to the relevant answer.
- Setup and installation questions are under Getting started.
- Trigger, output, and configuration questions are under Features.
- Models, images, capacity, and roles are under Configuration.
- Runs, recovery, security, and common failures are under Operations and Reference.
Frequently asked questions
What is the shortest path to a working setup?
Connect one repository, generate a runner token, run orchestrator install, configure one provider, start orchestrator run, and trigger one focused Code Agent or Issue Planner job.
Where does agent execution happen?
Inside CLI-managed Kubernetes workloads in the local runner environment.
Can I use different providers for different agents?
Yes. Provider, model, and supported thinking level are selected per agent.
How do I know an agent actually completed its job?
The run succeeds only after the runner verifies the expected GitHub artifact.
What remains in the hosted app?
Configuration, workspace data, jobs, events, sanitized provider metadata, status, usage telemetry, and verified outcome evidence.
What remains local?
Source checkouts, provider credentials, Docker and kind, persistent workspaces, and model execution.