slime-rl-training

Pass

Audited by Gen Agent Trust Hub on Jun 29, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides documentation for a legitimate machine learning framework (slime) developed by a recognized research group (THUDM).
  • [EXTERNAL_DOWNLOADS]: The skill instructs the user to download the framework from its official GitHub repository and pull the latest Docker image. These are standard procedures for using this software.
  • [COMMAND_EXECUTION]: The skill provides shell commands for environment setup and training execution. All commands are relevant to the primary purpose of RL post-training and do not show signs of malicious intent or unauthorized access.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 29, 2026, 12:54 AM
Security Audit — agent-trust-hub — slime-rl-training