safety-guard
Pass
Audited by Gen Agent Trust Hub on Aug 2, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements protective logic to monitor for high-risk terminal commands such as 'rm -rf', 'git push --force', and 'sudo rm'. These patterns are identified for the purpose of warning the user and requesting confirmation, rather than for malicious execution.
- [SAFE]: The 'Freeze Mode' functionality aims to restrict file writes to authorized directory trees (e.g., 'src/components/'), which follows the principle of least privilege during autonomous agent operation.
- [SAFE]: The skill uses a standard local path (~/.claude/safety-guard.log) for logging blocked actions, which is a benign practice for auditing and transparency.
- [SAFE]: There are no external network calls, obfuscated strings, or unverifiable third-party dependencies included in the skill definition.
Audit Metadata