autonomous-loops

Mostly a legitimate documentation skill for autonomous coding workflows, but it meaningfully increases operational risk because it normalizes unattended code changes, PR creation, CI-driven retries, and auto-merging. No clear credential theft or malicious exfiltration is present, yet the autonomous-action and prompt-injection surface make it medium/high risk in practice.