git-guardrails-claude-code
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a bash script intended to be executed by the agent platform as a hook. The script uses standard utilities (
jq,grep) to check command inputs against a predefined list of dangerous patterns. - [PERSISTENCE]: The skill modifies agent configuration files (
.claude/settings.jsonor~/.claude/settings.json) to register the hook. This ensures the guardrail remains active across different projects and sessions, which is the primary intended functionality of the skill.
Audit Metadata