nutrient-document-processing
Pass
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs users to run an MCP server using
npx -y @nutrient-sdk/dws-mcp-server, which downloads and executes the official SDK package from the npm registry. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted document content through OCR and text extraction, creating an ingestion point for potentially malicious instructions embedded in documents.
- Ingestion points: Documents (PDF, DOCX, images) uploaded via
curlor processed by the MCP server inSKILL.md. - Boundary markers: None identified in the provided instructions to separate data from instructions.
- Capability inventory: Performs network requests to the Nutrient API and writes processed output to the local file system.
- Sanitization: No explicit sanitization of extracted text or metadata is mentioned before being processed by the agent.
Audit Metadata