paper-lookup
Pass
Audited by Gen Agent Trust Hub on Aug 17, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [PROMPT_INJECTION]: The skill facilitates the ingestion of untrusted data from external academic databases, creating a surface for indirect prompt injection. Maliciously crafted paper abstracts or metadata could contain instructions intended to influence the agent's behavior. \n
- Ingestion points: API responses from PubMed, arXiv, Semantic Scholar, and other APIs referenced in the references directory. \n
- Boundary markers: The instructions lack specific guidance for the agent to treat fetched paper content as data only or to use delimiters to prevent command following. \n
- Capability inventory: The agent is empowered to use curl, WebFetch, and shell commands to retrieve and process this data. \n
- Sanitization: No instructions are provided to sanitize or validate the external content before processing it. \n- [COMMAND_EXECUTION]: The skill instructs the agent to use curl via bash or other shell environments as a primary or fallback method for making REST API calls. \n- [EXTERNAL_DOWNLOADS]: The skill performs network requests to multiple external domains belonging to academic services such as NCBI (NIH), arXiv (Cornell), and Crossref. These are well-known and reputable sources in the research community.
Audit Metadata