Agent Ops

Operating a running AI agent system — observing it, closing its improvement loop, governing its autonomy, keeping it reliable and bounded in cost — and the front-door for the agent skill family. Provenance lives in skill.json; this file is runtime routing only.

Produces: a change-plan.md (DO), an audit-report.md plus a findings-ledger + workflow-state when tracked (REVIEW), or a design-doc.md / refactor-runbook.md / explanation.md (DESIGN).

Core principle

Score on observed emission, not field presence; let autonomy run only behind a gate and a circuit-breaker. A system you cannot reconstruct from its traces, whose loop never closes, or that self-improves ungated fails at machine speed. agent-ops operates; it hands building off to the siblings below.

Activation