literature-review
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes scripts to fetch data from reputable academic sources including arXiv, Semantic Scholar, and OpenAlex. These are well-known services for scientific research, and the interactions are necessary for the skill's stated purpose.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external text from scientific papers, which creates an inherent surface for indirect prompt injection. This is mitigated by the research context and the skill's focus on grounded synthesis. * Ingestion points: The skill fetches paper summaries and full texts via the search scripts defined in SKILL.md and referenced in review-workflow.md. * Boundary markers: The instructions in dialogue-prompts.md and the SKILL.md synthesis steps require all claims to be grounded in the retrieved documents and cited, providing natural task-based boundaries. * Capability inventory: Capabilities include running local search scripts and generating text based on the results. * Sanitization: There is no explicit sanitization of external paper content described, as the agent is expected to summarize the technical findings.
Audit Metadata