autonomous-execution

The skill presents a coherent, self-contained autonomous execution loop designed for unattended development workflows. Its footprint—reading local state, performing a single task per iteration via subagents, validating, committing, and updating state—fits the stated purpose. Risk is modest and primarily organizational (governance of automated commits, ensuring proper handoff and re-evaluation) rather than technical (no external data flows or credential handling described). Monitor for governance enforcement (code reviews, protected branches) and ensure subagent permissions remain strictly scoped to the repository context.