Guides

Task-oriented pages. Each one uses only commands that exist in the CLI; see the CLI reference for every option.

• Write a canon: declare humans, hosts, agents and placements, check and sync them.
• Run the demo: the v1 scenario on the bench, step by step.
• Switch from a v0 setup: move a running team of agents to ameesh, one reversible step at a time.
• Cluster profile: one persistent agent per Kubernetes pod.
• Progress view: lots, agents, milestones and budget, as text, JSON or a standalone HTML page.
• Model catalogue: ameesh models.
• Review classes: ameesh review-class.
• Operate agents: session policy, list --json, cost per turn, gauges and balance, alerts, restart on a brief, interruption.
• Host ameesh-approve for a team: strict profile, local TLS, approve-check, database role, moving the service.