Version updated for https://github.com/provael/provael to version v0.37.0. This action is used across all versions by ? repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary Provael™ automates the red-teaming of Vision-Language-Action (VLA) robot policies through simulation. It evaluates policy success by running a series of attacks and measuring Attack Success Rate (ASR), providing an ASR-by-attack table, a pass/fail scorecard, and a SARIF report. The action is particularly useful for testing the robustness of VLA policies in various scenarios.

Version updated for https://github.com/provael/provael to version v0.36.2. This action is used across all versions by ? repositories. Action Type This is a Composite action. Go to the GitHub Marketplace to find the latest changes. Action Summary The Provael action automates the red-teaming of Vision-Language-Action (VLA) robot policies in simulation to measure their Attack Success Rate (ASR) by analyzing various attack techniques. It solves the problem of systematically testing VLA policies and generating comprehensive reports, including ASR statistics, pass/fail scores, and SARIF tags for rule evaluation. The action provides deterministic execution across tasks with reproducible results, making it a valuable tool for policy validation in simulation environments.