nutrient-document-processing
Pass
Audited by Gen Agent Trust Hub on Mar 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses
curlto interact with the official Nutrient DWS API (https://api.nutrient.io/build) for legitimate document processing operations such as format conversion and text extraction. - [EXTERNAL_DOWNLOADS]: The skill references the
@nutrient-sdk/dws-mcp-serverpackage, which is the official SDK package for the document processing service on the npm registry. - [REMOTE_CODE_EXECUTION]: The Model Context Protocol (MCP) configuration uses
npx -y @nutrient-sdk/dws-mcp-serverto execute the official service tool. This is an expected and safe integration method for well-known technology services. - [PROMPT_INJECTION]: The skill inherently processes untrusted document content (PDF, HTML) via OCR and text extraction, which creates a surface for indirect prompt injection if malicious instructions are embedded in the processed files.
- Ingestion points: Document files processed via the Nutrient DWS API in
SKILL.md. - Boundary markers: Absent; no specific delimiters or warnings for the agent are described in the instructions.
- Capability inventory: Shell access for
curl, network access to the service endpoint, and file system access for input and output operations. - Sanitization: Absent; the skill relies on the API service and does not specify additional local validation of extracted text.
Audit Metadata