extracting-mistral-ocr
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill interacts with the official Mistral AI API (api.mistral.ai) to perform OCR tasks and upload documents. It also supports processing documents from public URLs provided at runtime. These operations are consistent with the skill's stated purpose of document extraction and target a well-known service.
- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it ingests untrusted data from external PDFs and images.
- Ingestion points: External document content is processed by the
scripts/mistral_ocr_extract.pyscript via the--inputand--urlparameters. - Boundary markers: Not present. The script outputs the raw OCR-extracted Markdown and JSON data to the local filesystem without adding specific delimiters or warnings to the output files.
- Capability inventory: The script possesses the capability to perform network requests to Mistral AI and write files to any user-specified directory on the local disk.
- Sanitization: None. The script performs a direct mapping of the API response to local files. While this represents a standard injection surface for agents processing the resulting text, it is the intended primary function of the skill.
Audit Metadata