gh-cli
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides numerous shell command patterns for the GitHub CLI (gh) and git to automate repository management. All commands are standard and align with the stated purpose of automating GitHub operations.\n- [PROMPT_INJECTION]: The skill includes instructions that mandate user review and confirmation for all state-changing operations, effectively mitigating potential injection attempts. It also identifies a surface for indirect prompt injection via ingested GitHub data.\n
- Ingestion points: Commands like gh pr view, gh issue view, and gh run view --log-failed (SKILL.md, workflows/merge-pr.md) pull external content into the agent's context.\n
- Boundary markers: No explicit delimiter usage is specified for tool outputs.\n
- Capability inventory: The skill enables repository writes via gh pr create, gh pr merge, and gh issue comment.\n
- Sanitization: There is no automated sanitization of external GitHub data.
Audit Metadata