academic-paper-review
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze external, untrusted content from research papers and websites, creating a potential surface for indirect prompt injection.
- Ingestion points: Processes paper metadata and content from arXiv links, conference URLs, and uploaded PDFs as described in SKILL.md.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands within the research papers are specified.
- Capability inventory: The skill uses
web_fetchto retrieve external research data and saves results to a sandbox directory (/mnt/user-data/outputs/). - Sanitization: No sanitization or verification steps for the fetched or uploaded content are mentioned.
Audit Metadata