git-commit-messages
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWSAFE
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill is designed to process untrusted external data in the form of code diffs or change descriptions to generate commit messages. This creates a surface where an attacker could embed instructions within code comments to influence the agent's output. However, the risk is minimized because the skill explicitly requires displaying the generated message for user confirmation before any action is taken.
- [Prompt Injection] (SAFE): No attempts to bypass safety filters, override system instructions, or extract system prompts were detected in the instructions.
- [Data Exposure & Exfiltration] (SAFE): The skill does not access sensitive local files or perform network operations. All file paths mentioned (e.g., apps-server) are used as examples for commit scopes.
- [Unverifiable Dependencies & Remote Code Execution] (SAFE): No external scripts are downloaded, and no package managers (npm, pip, etc.) are invoked. The skill is purely instructional markdown.
- [Dynamic Execution] (SAFE): There is no use of eval, exec, or other dynamic code execution patterns.
Audit Metadata