git-guardrails-claude-code
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONPRIVILEGE_ESCALATION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes a shell script (
scripts/block-dangerous-git.sh) that is configured to run as aPreToolUsehook within the agent's environment to intercept commands. - [INDIRECT_PROMPT_INJECTION]: The safety script processes data from
tool_inputto identify blocked patterns. - Ingestion points:
scripts/block-dangerous-git.shreads command data from standard input. - Boundary markers: None present in the script logic.
- Capability inventory: The script is limited to checking input and exiting with a status code; it contains no file-write or network capabilities.
- Sanitization: Employs
jqfor structured data parsing andgrepfor pattern matching against a fixed blocklist. - [PRIVILEGE_ESCALATION]: The installation process requires using
chmod +xto make the hook script executable, which is the standard mechanism for enabling tool hooks in the supported platform.
Audit Metadata