pr
Pass
Audited by Gen Agent Trust Hub on Mar 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses git and the GitHub CLI (gh) to manage local branches, commit changes, and interact with remote repositories. These operations are essential to the skill's stated purpose and use standard system utilities.\n- [EXTERNAL_DOWNLOADS]: Fetches pull request metadata and diffs from GitHub. These operations interact with a well-known and trusted service necessary for pull request management.\n- [DATA_EXPOSURE]: Includes a clear directive for the agent to avoid committing API keys or private content, which serves as a proactive measure against accidental credential exposure.\n- [PROMPT_INJECTION]: The skill processes untrusted data from the repository's git log and pull request body, creating a surface for indirect prompt injection. 1. Ingestion points: git log, gh pr view. 2. Boundary markers: Absent. 3. Capability inventory: git commit, git push, gh pr edit. 4. Sanitization: Absent. The risk is considered minimal given the structured nature of the workflow.
Audit Metadata