isaaclab-debugging-rl-training
Pass
Audited by Gen Agent Trust Hub on Aug 13, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides procedural guidance and checklists for diagnosing reward hacking, checkpoint mismatches, and training instabilities in RL environments.
- [SAFE]: No external network operations, remote code downloads, or exfiltration patterns were detected.
- [SAFE]: The instructions do not involve accessing sensitive system files, environment variables, or hardcoded credentials.
- [SAFE]: The shell command mentioned (
uv run --no-project python tools/skills/cli.py check) is a local validation script for maintainers and does not pose a security risk.
Audit Metadata