git-guardrails-claude-code
Pass
Audited by Gen Agent Trust Hub on Sep 28, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides a security feature intended to limit the agent's authority over destructive Git operations. No malicious patterns or data exfiltration vectors were detected.
- [COMMAND_EXECUTION]: The setup process involves making a script executable (
chmod +x) and configuring the application to run it as a hook. This behavior is transparent, documented, and restricted to the skill's stated purpose of providing command guardrails. - [INDIRECT_PROMPT_INJECTION]: The provided script (
scripts/block-dangerous-git.sh) ingests untrusted tool execution data via stdin to perform command matching. - Ingestion points: Reads JSON input from stdin in
scripts/block-dangerous-git.sh. - Boundary markers: Uses standard JSON parsing via
jqto isolate the command string. - Capability inventory: Limited to string matching (
grep) and exiting with specific codes to signal the agent platform. - Sanitization: Employs
jqfor structured data extraction andgrepfor pattern identification, preventing command content from being executed by the hook itself.
Audit Metadata