scholar-slides
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python and Node.js scripts as part of its core functionality. These scripts handle PDF parsing (
PyMuPDF), chart generation (matplotlib), and high-fidelity rendering of slides and equations usingPlaywrightandKaTeXbased on local, generated HTML content. - [EXTERNAL_DOWNLOADS]: The skill is configured to download PDF papers from
arxiv.organd fetch bibliographic records fromcrossref.org. Both are established, well-known services within the academic domain and are used here for their intended primary purpose, qualifying as safe external references. - [PROMPT_INJECTION]: The skill ingests untrusted data from source PDFs during the 'digest' stage (
scripts/ingest_pdf.py), which represents an indirect prompt injection surface. This risk is effectively mitigated by the skill's design, which uses an intermediate structured 'digest' confirmed by the user (Checkpoint 1) before slide generation begins. Additionally, HTML escaping is used to prevent malicious content from being interpreted as code in the web-based slide output (scripts/lib/escape.mjs).
Audit Metadata