skills/openclaw/skills/agentguard/Gen Agent Trust Hub

agentguard

Pass

Audited by Gen Agent Trust Hub on Feb 21, 2026

Risk Level: SAFENO_CODE
Full Analysis
  • [SAFE] (SAFE): Analysis of the provided files confirms they are purely informational and metadata-based. No executable scripts, script generation, or prompt injections were found.
  • [COMMAND_EXECUTION] (SAFE): The skill references dangerous shell commands (e.g., fork bombs, recursive deletes) strictly within a security policy framework as patterns that should be denied by a monitor. It does not attempt to execute these commands.
  • [DATA_EXFILTRATION] (SAFE): The skill defines patterns for detecting sensitive information like private keys and API tokens to prevent their accidental or malicious transmission. No active data exfiltration or hardcoded credentials are present.
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill identifies high-risk domains and 'curl|bash' patterns as threats to be blocked, serving as a defensive reference rather than performing any downloads itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 21, 2026, 06:13 AM