resolving-merge-conflicts
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze data from external git metadata, which represents a potential injection surface.
- Ingestion points: The instructions direct the agent to read commit messages, PRs, and original issues to understand the intent of changes (SKILL.md).
- Boundary markers: There are no boundary markers or instructions to treat external data as untrusted content.
- Capability inventory: The agent is authorized to run automated project checks (tests, linters) and stage/commit changes to the repository.
- Sanitization: No sanitization or validation logic is defined for the external text being processed.
Audit Metadata