zhuoyu-workshop
Pass
Audited by Gen Agent Trust Hub on Apr 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is centered on academic integrity and honest project representation. It defines clear boundaries against fabricating awards, metrics, or research outcomes.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface: The skill is designed to analyze untrusted materials such as source code and reports provided by the user. While it has capabilities for code editing and command execution, it mitigates risks through strict operating rules and a structured diagnostic framework. Ingestion points include repository files and logs specified in SKILL.md. Capabilities include code modifications and validation command execution defined in references/ai-coding-workflow.md. Boundary markers and explicit sanitization logic are not present, but the risk is managed by the skill's prescriptive workflow.
Audit Metadata