document-processing
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to extract text and structured data from external, potentially untrusted documents (PDFs, images, HTML) which may contain hidden instructions intended to influence the agent's behavior.
- Ingestion points: Files are ingested through multiple tools including pdftotext, pymupdf, and the Mistral OCR API as described in SKILL.md.
- Boundary markers: The instructions do not define clear delimiters or 'ignore instructions' prompts to separate extracted content from the agent's core logic.
- Capability inventory: The skill has access to the file system (reading/writing files), network capabilities (via httpx for Mistral API), and shell command execution (via pdftotext, tesseract, and pandoc).
- Sanitization: No sanitization or validation of the extracted document content is performed before it is presented to the agent.
Audit Metadata