git
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes Bash to execute standard Git operations (add, commit, push, etc.), which is expected for its stated purpose.
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes repository context and file content to generate intelligent commit messages. This introduces a surface area where adversarial content within a processed repository could attempt to influence the agent's behavior.
- Ingestion points: Repository context and files accessed via 'Read' integration described in SKILL.md.
- Boundary markers: None explicitly defined in the provided instructions to isolate untrusted repository data from the system prompt.
- Capability inventory: Execution of Git commands and general Bash access as defined in the 'Claude Code Integration' section.
- Sanitization: No explicit sanitization or filtering of repository content is mentioned before processing for commit message generation.
Audit Metadata