agent-guardrails
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill provides Bash scripts (
guard-bash.sh,lint-edited-file.sh,stop-verify.sh) designed to be executed by the AI agent's hook system. These scripts perform local repository checks, such as blocking specific shell command patterns and running linting tools. - [EXTERNAL_DOWNLOADS]: The documentation references external security resources and curated skill lists from established organizations including Anthropic, Trail of Bits, and Cisco to assist users in vetting third-party extensions.
- [DATA_EXFILTRATION]: An audit script
scripts/check_guardrails.pyperforms read-only inspection of local configuration files (e.g.,.claude/settings.json,AGENTS.md) to verify the integrity of the guardrail setup without transmitting data externally.
Audit Metadata