wos-download

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill parses external web content from academic publisher sites, which presents a risk of indirect prompt injection. * Ingestion points: Data is extracted from web pages in SKILL.md during Step 2 and Step 4 using document.querySelector. * Boundary markers: There are no explicit markers or instructions to isolate untrusted web content from the agent's logic. * Capability inventory: The skill has the ability to navigate pages, click elements, and execute browser scripts. * Sanitization: No evidence of data sanitization or validation is present for the ingested text.
  • [COMMAND_EXECUTION]: The skill uses browser automation commands to navigate websites and interact with page elements, including an initialization script that modifies the browser's navigator.webdriver property to bypass automated bot detection.
  • [EXTERNAL_DOWNLOADS]: The skill triggers the download of PDF files from external publisher domains to the user's local system based on findings from the Web of Science database.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 09:38 AM