resolve-conflicts
Pass
Audited by Gen Agent Trust Hub on Jul 30, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses standard, read-only git commands such as status and branch to gather local environment context upon loading. These operations are benign and do not involve sensitive data access.
- [COMMAND_EXECUTION]: The skill uses the bash shell to perform git operations and execute project verification scripts (build, test, lint). All command patterns are standard for development workflows and stay within the expected scope of a conflict resolution tool.
- [PROMPT_INJECTION]: The skill processes external data from git history and GitHub PR descriptions, which constitutes a potential indirect prompt injection surface.
- Ingestion points: Git commit messages, PR titles/bodies (via gh), and file contents/diffs.
- Boundary markers: None explicitly defined, but the instructions enforce a manual 'intent recovery' step that requires the agent to summarize each side's changes before editing.
- Capability inventory: Shell access for git commands and project-specific validation tools.
- Sanitization: The skill relies on agent-led history analysis and the project's existing build/test gates to ensure the integrity of the resulting code.
Audit Metadata