devextreme-form

Pass

Audited by Gen Agent Trust Hub on Aug 18, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The 'Smart Paste' functionality described in references/ai-smart-paste.md allows the form to ingest unstructured text from a user's clipboard and process it via an LLM. This creates a surface for indirect prompt injection where malicious content in the pasted text could attempt to influence the AI's data extraction behavior.
  • Ingestion points: Unstructured text passed to the smartPaste method or the smartPaste button item in references/ai-smart-paste.md.
  • Boundary markers: None explicitly defined in the client-side implementation examples.
  • Capability inventory: Modifies form field values (e.aiResult) and performs network requests (fetch) to a developer-defined AI backend.
  • Sanitization: The skill documentation notes that the result is returned as a plain object, but the input text itself is not sanitized as it is intended for AI processing.
  • [EXTERNAL_DOWNLOADS]: The documentation in references/ai-smart-paste.md provides instructions for loading the dx.ai-integration.js library via a script tag targeting Cloudflare's CDN. This is a reference to a well-known service for delivering official component assets.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 18, 2026, 01:48 PM
Security Audit — agent-trust-hub — devextreme-form