aoti-debug
Pass
Audited by Gen Agent Trust Hub on Jun 28, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill metadata refers to the official PyTorch repository, which is a trusted and well-known source for development tools. The documentation and guides provided are consistent with official project practices and intended for use in professional development environments.
- [COMMAND_EXECUTION]: The skill instructs the use of common development tools like
grepto search generated source code for debugging patterns. It also suggests setting standard framework-specific environment variables (e.g.,CUDA_LAUNCH_BLOCKING=1,AOTI_RUNTIME_CHECK_INPUTS=1) to enable debugging features within the PyTorch ecosystem. - [SAFE]: No malicious patterns, such as prompt injection, code obfuscation, or unauthorized data exfiltration, were identified. The skill's primary function is to assist in legitimate software development and troubleshooting of deep learning models.
Audit Metadata