ascend-inference-repos-copilot
Pass
Audited by Gen Agent Trust Hub on Apr 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for technical Q&A and documentation retrieval for specific open-source projects related to Ascend NPU inference. It follows legitimate patterns for intent identification and repository routing.
- [DATA_EXPOSURE]: The instructions ask the user for system configuration details (such as Ascend chip models and CANN versions) strictly for troubleshooting purposes. This is standard practice for technical support and does not involve harvesting sensitive personal or credential data.
- [REMOTE_CODE_EXECUTION]: No evidence of arbitrary command execution, remote script downloads (e.g., curl|bash), or unauthorized package installations was found.
- [PROMPT_INJECTION]: The skill does not contain instructions to override AI safety filters or extract system prompts. It explicitly prohibits the fabrication of technical details.
- [INDIRECT_PROMPT_INJECTION]: While the skill retrieves content from external repositories based on user input, it includes safety-enhancing instructions such as query optimization, disambiguation protocols, and verification against source code to maintain accuracy and prevent the propagation of misleading information.
Audit Metadata