conflict-prevention-engine
Pass
Audited by Gen Agent Trust Hub on Sep 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted data such as Pull Request metadata, branch differences, and commit messages (referenced in
analyze-pr,analyze-merge, and CI/CD integration examples inSKILL.md). - Ingestion points: Data entering the agent context includes Pull Request content and branch metadata processed by
tools/conflict_prevention_engine.py. - Boundary markers: The documentation and example code do not specify the use of delimiters or 'ignore embedded instructions' warnings when processing this external content.
- Capability inventory: The skill uses subprocess calls to execute Python tools and integrates with CI/CD workflows (GitHub Actions) which could be influenced by injected instructions.
- Sanitization: There is no evidence in the provided templates of sanitization, escaping, or strict schema validation for the data ingested from external version control sources.
Audit Metadata