nutrient-document-processing
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions provide examples for using
curlto interact with the Nutrient DWS API and suggest usingnpxto execute the@nutrient-sdk/dws-mcp-serverpackage. These are standard operations for accessing the document processing service. - [EXTERNAL_DOWNLOADS]: The skill references the official Nutrient SDK package on NPM and utilizes the
api.nutrient.iodomain for processing requests. These sources are recognized as established services for the skill's documented functionality. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, potentially untrusted documents (such as PDFs and DOCX files) to extract text, data, or perform OCR. This establishes an indirect injection surface if the agent subsequently processes the extracted content. \n
- Ingestion points: Local document files uploaded via
curlmultipart requests or the MCP server.\n - Boundary markers: None present in the instructional examples to distinguish extracted content from agent commands.\n
- Capability inventory: Shell access (via
curl), remote package execution (npx), and file system writes for outputting processed documents.\n - Sanitization: The instructions do not define sanitization or filtering steps for the data returned from the processing API.
Audit Metadata