fix-merge-conflicts
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process code containing merge conflict markers. If the conflicts originate from an untrusted source (e.g., a public pull request), the agent may be exposed to malicious instructions or code designed to exploit the build or test environment.
- Ingestion points: Local files identified by
git statuscontaining conflict markers. - Boundary markers: Absent; the skill instructs the agent to edit the files directly to resolve conflicts.
- Capability inventory: Uses local build, lint, and test tools as well as package managers (e.g., npm, pip) to validate resolutions.
- Sanitization: None; relies on the LLM's reasoning to perform 'correctness-first' edits.
Audit Metadata