model-capability-router

Pass

Audited by Gen Agent Trust Hub on Aug 16, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The scripts/route.py and scripts/test_route.py files use subprocess.run to execute Python scripts for routing validation and unit testing. These executions are limited to the skill's own local scripts using the current environment's interpreter.
  • [REMOTE_CODE_EXECUTION]: No remote code execution was found. All scripts are local to the skill directory, and there are no instances of curl | bash or similar patterns involving external untrusted sources.
  • [DATA_EXFILTRATION]: No data exfiltration patterns were detected. The skill does not access sensitive directories (e.g., .ssh, .aws) or perform outbound network requests to unknown domains.
  • [PROMPT_INJECTION]: The skill instructions define internal role boundaries and task hierarchies. There are no attempts to bypass safety filters or override global system instructions.
  • [CREDENTIALS_UNSAFE]: No hardcoded credentials or secrets were found in the provided configuration files or scripts.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 16, 2026, 01:23 AM
Security Audit — agent-trust-hub — model-capability-router