query
Pass
Audited by Gen Agent Trust Hub on Jul 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The wrapper script
bin/querydownloads a platform-specific binary from the vendor's CDN athttps://agent-cdn.nutrient.io/pdf-to-markdown/. This is performed during the first run or during periodic update checks. - [COMMAND_EXECUTION]: The script executes the downloaded binary to perform search operations. It passes query parameters and file paths directly to the tool.
- [PROMPT_INJECTION]: The skill ingests untrusted text from documents for searching. While results are restricted to specific line windows, the lack of sanitization or boundary markers for the extracted text presents a surface for indirect prompt injection where malicious content in the document could attempt to influence the agent.
Audit Metadata