commit
Pass
Audited by Gen Agent Trust Hub on Jun 22, 2026
Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
- [SAFE]: The skill performs standard local git operations as part of its primary functionality.
- [DATA_EXFILTRATION]: The prompt includes explicit instructions to exclude sensitive files such as .env when staging changes, which is a key security measure against data exposure.
- [COMMAND_EXECUTION]: Command usage is restricted to standard git utilities (status, diff, log, add, commit) used locally.
- [PROMPT_INJECTION]: While the skill analyzes external data (git diffs and conversation history) which could contain indirect prompt injections, the requirement for a user confirmation step before executing the final commit plan effectively mitigates this risk.
Audit Metadata