skills/syncfusion/angular-ui-components-skills/syncfusion-angular-inline-ai-assist/Gen Agent Trust Hub
syncfusion-angular-inline-ai-assist
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: Documentation examples demonstrate a pattern where untrusted AI-generated content is interpolated into the application's DOM without sanitization. \n
- Ingestion points: Handlers for promptRequest and itemSelect in SKILL.md, references/getting-started.md, references/core-configuration.md, and references/events-and-methods.md ingest external AI responses. \n
- Boundary markers: The examples lack specific markers or instructions to the AI to ignore embedded commands. \n
- Capability inventory: The component is designed to incorporate AI content into the user interface and editable text areas. \n
- Sanitization: Multiple code snippets use innerHTML to update the DOM (e.g., editable.innerHTML = lastResponse) without demonstrating sanitization, creating a potential cross-site scripting (XSS) surface. \n- [EXTERNAL_DOWNLOADS]: The skill documentation includes examples of fetching data from external services and installing packages. \n
- Evidence: The references/getting-started.md file provides instructions to install official Syncfusion packages (e.g., @syncfusion/ej2-angular-interactive-chat) via npm. The references/core-configuration.md file includes an example of fetching streaming responses from OpenAI's API (api.openai.com).
Audit Metadata