parity-testing

Pass

Audited by Gen Agent Trust Hub on Apr 19, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill contains instructional content for developers working on model conversion. No malicious patterns such as prompt injection, data exfiltration, or obfuscation were detected.
  • [COMMAND_EXECUTION]: The skill provides documentation on how to use standard command-line tools (e.g., uv, torch.distributed.run) to execute local verification scripts located in the examples/conversion/ directory. These commands are intended for local development and debugging workflows.
  • [DATA_EXPOSURE_AND_EXFILTRATION]: No evidence of sensitive data access or exfiltration. The placeholders used in command examples (e.g., <org>/<model>) are standard for identifying public model repositories.
  • [DYNAMIC_EXECUTION]: Includes Python code snippets for tensor and state-dict comparison. These are static utility functions for numerical verification and do not involve unsafe execution of untrusted code.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 19, 2026, 08:13 AM
Security Audit — agent-trust-hub — parity-testing