k2f-academic-paper
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, such as user notes, preprints, and PDFs, to generate academic manuscripts. \n
- Ingestion points: User-supplied materials including research notes, PDFs, and web text are processed within the workflow. \n
- Boundary markers: The skill contains explicit instructions to treat user evidence as data for summarization only, warning not to let it override the defined scope. \n
- Capability inventory: The skill creates .K2F packages and triggers exports to multiple file formats (PDF, DOCX, etc.) via the k2f tool. \n
- Sanitization: No automated sanitization is described; the skill relies on a structured blueprinting workflow. \n- [EXTERNAL_DOWNLOADS]: The instructions direct the agent to install a core component (suzheng/k2f) using the npx skills add command. As this resource is from the skill's own author, it represents vendor-intended functionality.
Audit Metadata