ralph-github-start-loop

SUSPICIOUS. The core GitHub automation behavior matches the stated purpose and data stays within GitHub, but the skill enables high-impact autonomous repository actions and relies on a third-party gh extension with weaker trust guarantees than first-party gh features. This is not confirmed malware, but it is a medium-to-high risk workflow skill.

This fragment appears to be an autonomous GitHub development orchestrator rather than overt malware. The primary risks are unrestricted Claude execution via --dangerously-skip-permissions, prompt injection through PRD/issue/repository content, and automatic git push and issue closure based on minimally validated commit text. It should be isolated, run with least-privilege credentials, validate diffs and tests, and require approval before publishing or closing issues.