address-github-comments
Pass
Audited by Gen Agent Trust Hub on Apr 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes external data from GitHub Pull Request comments, which creates a surface for indirect prompt injection where a contributor might attempt to influence the agent's behavior.\n
- Ingestion points: PR comments are retrieved via the
gh pr view --commentscommand inSKILL.md.\n - Boundary markers: The workflow explicitly instructs the agent to 'Wait for user confirmation' before addressing comments, providing a human-in-the-loop mitigation.\n
- Capability inventory: The agent is expected to modify source code and execute
ghCLI commands based on the feedback.\n - Sanitization: There is no explicit technical sanitization of the comment content described in the workflow.\n- [COMMAND_EXECUTION]: The skill utilizes the official GitHub CLI (
gh) for authenticating, viewing pull request comments, and posting responses. These operations are standard for the intended developer use case and involve interactions with a well-known service (GitHub).
Audit Metadata