paper-polisher
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external content from paper drafts and method explanations, creating a surface for indirect prompt injection where instructions embedded in the processed data could attempt to influence the agent's behavior. 1. Ingestion points: Reads draft files from paper/sections/.md, paper/sections/.tex, methods/Qx/qx_final_method_explanation.md, and results/Qx/reports/qx_final_result_analysis.md. 2. Boundary markers (absent): The skill does not define explicit delimiters or boundary markers to isolate untrusted data during the polishing process. 3. Capability inventory: The skill performs file reading and text generation. No network or arbitrary command execution capabilities are utilized. 4. Sanitization (absent): The instructions do not specify sanitization or filtering logic for the content being processed.
- [EXTERNAL_DOWNLOADS]: The skill references an external GitHub repository (github.com/Yuan1z0825/nature-skills) as a design inspiration. This is a documentation reference only; the skill does not download or execute code from this URL.
Audit Metadata