skills/cosmix/loom/loom-refactoring/Gen Agent Trust Hub

loom-refactoring

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No prompt injection patterns or attempts to bypass safety filters were detected in the instructions.
  • [COMMAND_EXECUTION]: The skill utilizes shell commands through ripgrep for code analysis and mentions executing local test suites. These actions are standard for the task of refactoring and occur within the local development environment.
  • [DATA_EXFILTRATION]: No network operations or instructions to exfiltrate data were found. The skill operates entirely within the context of the local codebase.
  • [REMOTE_CODE_EXECUTION]: The skill does not download or execute untrusted code from external sources.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 03:57 AM
Security Audit — agent-trust-hub — loom-refactoring