academic-figure-generation
Fail
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The
scripts/generate.pyscript usessys.path.append()to dynamically include a user-specified directory in the Python module search path. It then imports numerous classes and agents from this directory at runtime. Dynamic loading of code from computed paths is a high-risk pattern. - [REMOTE_CODE_EXECUTION]: The skill's core functionality relies on executing code downloaded from an external, unvetted GitHub repository (
dwzhu-pku/PaperBanana). This establishes a mechanism for running arbitrary third-party logic that has not been audited for safety. - [EXTERNAL_DOWNLOADS]: The skill documentation provides instructions to clone the PaperBanana repository and install its dependencies. While GitHub is a well-known service, the specific repository is an unvetted third-party source.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process arbitrary method descriptions and captions through a multi-agent LLM pipeline. This creates an attack surface where malicious instructions embedded in the input text could influence the behavior of the underlying agents.
Recommendations
- AI detected serious security threats
Audit Metadata