git-guardrails-claude-code

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEPERSISTENCECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PERSISTENCE]: The skill provides instructions to modify the agent's global or project-specific configuration files (settings.json) to install hooks that persist across sessions.
  • [COMMAND_EXECUTION]: The hook configuration utilizes a Bash matcher to execute a shell script, intercepting and controlling the execution of Git commands.
  • [PRIVILEGE_ESCALATION]: The setup process requires the user to grant execution permissions to the hook script using the chmod +x command.
  • [INDIRECT_PROMPT_INJECTION]: The safety hook processes raw tool inputs, which represents a vulnerability surface for indirect injection attacks.
  • Ingestion points: PreToolUse hook intercepting tool_input (SKILL.md)
  • Boundary markers: No delimiters or warnings implemented in the instructions
  • Capability inventory: Shell script execution via hook mechanism
  • Sanitization: None described in the setup process; implementation details are limited to the referenced but external hook script.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 10:32 PM
Security Audit — agent-trust-hub — git-guardrails-claude-code