pm-export
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local Python script (push_to_zotero.py) to process research data and communicate with the Zotero desktop application via its local connector API.
- [EXTERNAL_DOWNLOADS]: The skill fetches bibliographic metadata and full-text PDF files from official, trusted NCBI domains (eutils.ncbi.nlm.nih.gov and www.ncbi.nlm.nih.gov).
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests paper metadata from external scientific databases. 1. Ingestion points: Bibliographic data is fetched via the evaluate_script tool calling NCBI E-utilities, and PDF files are downloaded by the Python script. 2. Boundary markers: No explicit delimiters or instructions are used to separate the external data from the agent's context. 3. Capability inventory: The skill can execute browser-based JavaScript, run a local Python script, and perform local network operations to interface with Zotero. 4. Sanitization: The skill parses structured JSON and XML data but does not perform specialized sanitization on text fields like abstracts or titles before processing.
Audit Metadata