git-resolve-conflicts
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is instructed to recover change intent by reading git commit messages and following references to external trackers or pull requests. This ingestion of untrusted data creates a surface for indirect prompt injection where malicious instructions embedded in the history could influence the agent. * Ingestion points: git log output and external issue tracker/pull request content (SKILL.md). * Boundary markers: The skill does not employ delimiters or specific instructions to ignore embedded commands within the ingested text. * Capability inventory: Subprocess calls for git operations and project-specific build/test scripts (SKILL.md). * Sanitization: No sanitization or validation is applied to the recovered intent text.
- [COMMAND_EXECUTION]: The agent is directed to discover and run build and test commands defined in project metadata files such as package.json, Makefile, and pyproject.toml. This involves the dynamic execution of arbitrary shell commands defined within the local repository environment to verify the integrity of the conflict resolution.
Audit Metadata