cutedsl-kernel-integration
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill serves as a technical integration guide for cuDNN Frontend developers. It contains no instructions that attempt to bypass AI safety guidelines or override agent behavior.
- [SAFE]: There are no instances of hardcoded credentials, sensitive file access, or unauthorized network operations. References to file paths are limited to local repository structures typical for a Python project.
- [SAFE]: The skill does not employ any obfuscation techniques such as Base64 encoding, zero-width characters, or homoglyph substitutions.
- [SAFE]: Dependency management follows standard practices, specifically referencing an existing
cutedsloptional dependency in the project's configuration. No suspicious remote code execution patterns were identified. - [SAFE]: The instructions do not involve privilege escalation, persistence mechanisms, or dynamic execution of untrusted code. All described workflows (class extensions, wrapper implementations, and unit testing) are standard for the stated development task.
Audit Metadata