resolving-merge-conflicts
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions follow standard software development best practices for conflict resolution and project testing.
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from git history and issue trackers, which could potentially contain malicious instructions intended to influence the agent's behavior.
- Ingestion points:
SKILL.md(Step 2) instructs the agent to read commit messages, PRs, and originating issues/tickets. - Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are present.
- Capability inventory:
SKILL.md(Steps 4 and 5) allows the agent to execute automated checks (tests, typechecks) and perform git operations (staging, committing). - Sanitization: No sanitization or validation logic is specified for the external content being read.
Audit Metadata