enterprise-agent-ops

This skill provides operational controls for cloud-hosted or continuously running agent systems. Use it when you need runtime lifecycle management, observability, safety controls, and change management beyond single CLI sessions.

• Manage runtime lifecycle (start, pause, stop, restart) for agent workloads
• Implement observability through logs, metrics, and traces
• Enforce safety controls including scopes, permissions, and kill switches
• Execute change management with rollout, rollback, and audit capabilities
• Deploy immutable artifacts with least-privilege credentials and environment-level secret injection
• Track key metrics: success rate, mean retries, time to recovery, cost per task, and failure distribution

1. 1.Set up immutable deployment artifacts and configure least-privilege credentials for your agent workload
2. 2.Inject environment-level secrets and establish hard timeout and retry budgets
3. 3.Configure observability collection for logs, metrics, and traces from your running agents
4. 4.Establish baseline safety controls including scopes, permissions, and kill switches
5. 5.When failures occur, freeze new rollouts and capture representative traces to isolate the failing route
6. 6.Apply the smallest safe patch, run regression and security checks, then resume rollout gradually