pr
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external and untrusted data from pull request reviews and CI logs, creating an attack surface where malicious feedback could influence agent actions.
- Ingestion points: Pull request status, review comments, CI state, and conflict markers as specified in the 'Steps' section of
SKILL.md. - Boundary markers: There are no instructions to use delimiters or explicitly ignore embedded instructions within the ingested PR data.
- Capability inventory: The skill instructions include 'Make changes', 'Rebase or merge', and 'Re-run CI' (
SKILL.md), which involve file system writes and the execution of git or CI tooling. - Sanitization: No sanitization, filtering, or validation of the external comments or CI output is defined before the agent acts upon the content.
Audit Metadata