Every prompt, tested and categorized.

Write a production system prompt: role, scope, tools, rules, failure behaviour.

Plan the eval gate, canary and rollback for a prompt or model change.

A blameless postmortem with timeline, root cause and action items.

Pick the few service-level indicators and targets that matter for your agent.

Imagine the launch failed; list how and what guard stops each failure.

Pressure-test whether your task really needs multiple agents.