Evaluating Paper Relevance
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes curl to interact with established academic and scientific repositories, including NCBI PubMed Central, EBI ChEMBL, and Unpaywall, for retrieving paper metadata and full-text content.
- [COMMAND_EXECUTION]: Uses standard command-line tools such as curl for network data retrieval and grep for searching patterns within the retrieved research documents.
- [PROMPT_INJECTION]: The skill ingests untrusted data in the form of paper abstracts and full text. This exposure is managed through a structured scoring system and specific extraction templates, reducing the risk associated with indirect prompt injection.
- [DATA_EXFILTRATION]: Requests a user-provided email address specifically for Unpaywall API access, which is a standard authentication requirement for that service and is used legitimately within the research workflow.
Audit Metadata