cybersecurity-owasp-trainer

To train cybersecurity models using OWASP guidelines, this Skill orchestrates an end-to-end training loop using TRL/GRPO and Unsloth. It provides a structured framework that preserves verifier integrity and reward signals throughout the model training cycle.

OpenEnv guides secure code generation workflows by orchestrating deterministic training steps with clear gating. It ensures safe, structured experimentation for CyberSecurity_OWASP model runs while maintaining verifier integrity and consistent reward signals.

Yes, you can generate rollout artifacts and run baseline-vs-trained evaluations. The Skill coordinates these evaluations using OpenEnv and TRL to track progress and demonstrate trained model performance improvements across secure seeds.

Setting up Modal for scalable cybersecurity reinforcement learning experiments involves running preflight checks, configuring caching strategies, and executing safe smoke runs. The Skill provides explicit guidance for these specific scalable experimentation steps.

Yes, the workflow supports Trackio metrics and dashboard generation for hackathon demos. It provides specific guidance for creating metrics, dashboards, and demo artifacts to document baseline-vs-trained performance evaluation evidence.