nudocs
Pass
Audited by Gen Agent Trust Hub on Mar 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Installs the
@nutrient-sdk/nudocs-clipackage from npm. This is the official CLI tool for the Nudocs service, provided by a well-known technology vendor (Nutrient/PSPDFKit). - [COMMAND_EXECUTION]: Executes the
nudocscommand-line tool to upload, list, and download documents. The documentation inreferences/formats.mdcontains examples using agimmecommand, which is likely a documentation alias or error and does not impact security. - [DATA_EXFILTRATION]: Uploads user-provided document content to
https://nudocs.aifor editing and sharing. This is the core functionality of the skill and is directed to the official service domain. - [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface. * Ingestion points: Files downloaded via
nudocs pulland subsequently read by the agent. * Boundary markers: None identified in the skill instructions to delimit external content. * Capability inventory: Subprocess execution of thenudocsCLI and local file system access. * Sanitization: No explicit sanitization or filtering of downloaded content is performed before presentation to the agent.
Audit Metadata