git
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed as a read-only assistant for Git operations. It explicitly instructs the agent to 'NEVER EXECUTE write commands' and to only provide commands in code blocks for the user to manually run, which is a significant security control.
- [COMMAND_EXECUTION]: While the skill mentions various Git commands (including destructive ones like 'reset --hard' and 'push -f'), these are documented as templates for the user. The internal security protocol restricts the agent's autonomous behavior to read-only commands like 'status', 'log', and 'diff'.
- [DATA_EXFILTRATION]: No evidence of network operations, hardcoded credentials, or access to sensitive configuration files (e.g., .ssh/config, .aws/credentials) was found.
Audit Metadata