OverviewHistoryStatsSecurity
Security
Independent checks from skills.sh's audit partners.
Independent checks from skills.sh's audit partners.
Warn4 providerslatest audit Apr 16, 2026
Independent checks from skills.sh's audit partners.
This skill provides guidance for training LLMs with the verl reinforcement learning framework. It includes instructions for distributed training workflows using shell commands and cluster management via Ray. It also allows for dynamic loading of custom Python reward functions from local paths and processes large datasets, which may present surfaces for code execution and indirect prompt injection if untrusted inputs are used. External resources are sourced from the framework's official repositories.
Detected behaviors
No alerts
1 issue
Score: 93/100 · 2 sections analyzed