review-pr-in-lane
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon data from external, untrusted sources such as GitHub issue comments and Pull Request content.
- Ingestion points: The skill uses the
Monitortool to poll GitHub issue comments every 5 minutes andReadNotificationsto receive updates. It also fetches PR data usinggh pr view. - Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat the content of the issues or PRs as data rather than instructions.
- Capability inventory: The skill utilizes
ListAgents,Monitor,ReadNotifications, andghCLI tools, and transitions to worktree-based code review which likely involves file system operations. - Sanitization: While the skill performs a logical check to ensure the PR's
headRefNamematches the expected integration branch, it does not explicitly sanitize the external text content before processing. - [DATA_EXFILTRATION]: During the 'check-in' phase, the skill directs the agent to post its internal session name (obtained via
ListAgents) as a public comment on a GitHub issue. This behavior discloses agent-specific metadata to external parties or other users monitoring the repository.
Audit Metadata