enterprise-agent-ops

Operational controls for long-lived agent workloads with observability, security, and lifecycle management. Use this skill for cloud-hosted or continuously running agent systems that need operational controls beyond single CLI sessions. When failure spikes: This skill pairs with: - Covers four operational domains: runtime lifecycle (start, pause, stop, restart), observability (logs, metrics, traces), safety controls (scopes, permissions, kill switches), and change management (rollout, rollback, audit) - Enforces baseline controls including immutable deployment artifacts, least-privilege credentials, environment-level secret injection, and hard timeouts with retry budgets - Tracks key metrics: success rate, mean retries per task, time to recovery, cost per task, and failure class distribution - Includes structured incident response: freeze rollouts, capture traces, isolate failures, patch safely, run regression and security checks, then resume gradually - Integrates with PM2, systemd, container orchestrators, and CI/CD gates - runtime lifecycle (start, pause, stop, restart) - observability (logs, metrics, traces) - safety controls (scopes, permissions, kill switches) - change management (rollout, rollback, audit) - immutable deployment artifacts - least-privilege credentials - environment-level secret injection - hard timeout and retry budgets - audit log for high-risk actions...