nutrient-document-processing
Pass
Audited by Gen Agent Trust Hub on Mar 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructions suggest the use of an alternative MCP server using
npx -y @nutrient-sdk/dws-mcp-server, which downloads a package from the npm registry during execution. - [REMOTE_CODE_EXECUTION]: The
npxcommand is used to download and execute the@nutrient-sdk/dws-mcp-serverpackage from npm. - [COMMAND_EXECUTION]: The skill provides multiple examples using the
curlutility to send local files to the Nutrient API for processing. - [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes external, potentially untrusted documents (PDF, DOCX, etc.).
- Ingestion points: Local files and HTML content passed as multipart form data to the API via
curlor handled by the MCP server (SKILL.md). - Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands provided in the skill guidelines.
- Capability inventory: The skill utilizes shell commands (
curl,npx) and interacts with external API endpoints (SKILL.md). - Sanitization: No evidence of sanitization, validation, or escaping of document content is present in the skill definition.
Audit Metadata