literature-search
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials or access to sensitive file paths (~/.ssh, .env, etc.) were found. The skill interacts with public academic databases.
- [Remote Code Execution] (SAFE): No remote script downloads or unverified package installations (pip/npm) are present.
- [Indirect Prompt Injection] (LOW): The skill ingests external data from PubMed (titles and abstracts). While this represents a theoretical attack surface for indirect injection, the skill lacks high-risk capabilities like shell execution or arbitrary file writing.
- Ingestion points: search_literature and fetch_article_details (external PubMed content).
- Boundary markers: Not explicitly defined in instructions.
- Capability inventory: Zotero import and data merging.
- Sanitization: None described in the skill markdown.
- [No Code] (SAFE): The skill consists of a markdown file describing tool interactions and does not include any scripts or binary components.
Audit Metadata