nature-paper2ppt
Pass
Audited by Gen Agent Trust Hub on Oct 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of scientific papers and notes, creating an indirect prompt injection surface.
- Ingestion points: The skill accepts paper PDFs, preprints, article text, and reading notes as inputs (referenced in SKILL.md and static/core/workflow.md).
- Boundary markers: There are no explicit delimiters or boundary markers mentioned to isolate the untrusted document content from the agent's instructions.
- Capability inventory: The toolchain utilizes python-pptx for file writing and PyMuPDF for file reading/rendering (static/core/toolchain.md).
- Sanitization: No specific sanitization or filtering for adversarial content is implemented, although a self-review loop for scientific accuracy is described in references/self-review.md.
Audit Metadata