Documentation
Run agents. Read the manual.
The full FerrFleet documentation is being assembled. In the meantime, the most useful starting points are below; each links to the GitHub repository where the source of truth currently lives.
01
Getting started
Install the CLI, point it at your cluster, dispatch your first run. The five-minute path from nothing to a live agent on the dashboard.
02
Agent catalog format
The YAML contract that defines an agent: prompts, tool configs, model cascade, sandbox policy, schedule. Field-by-field reference with sane defaults.
03
Runs API
Dispatch a run, fetch the transcript, replay with a different model, stream live tokens over SSE. REST endpoints, auth, error shapes, rate limits.
04
Self-host Helm chart
Run FerrFleet on your Kubernetes cluster. Postgres, Redis, object storage, Vault Agent for token injection, with the values.yaml reference, upgrade path and backup story.
05
Self-hosted runner pools
Run your agents on machines you host, with your own Anthropic API key. Create a pool, start runners with the Helm chart, Docker Compose or one container, and point an agent at it.