openui-forge-langchain
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: Secret Management: The skill correctly advises users to manage API credentials using environment variables (OPENAI_API_KEY, ANTHROPIC_API_KEY) rather than hardcoding them into scripts.
- [SAFE]: Dependency Integrity: All listed dependencies belong to the established LangChain ecosystem or the author's @openuidev namespace, aligning with the skill's generative UI functionality.
- [INDIRECT_PROMPT_INJECTION]: The skill implements a chat interface that ingests user-provided text to drive UI generation.
- Ingestion points: The POST handler in app/api/chat/route.ts accepts a JSON body containing user messages.
- Boundary markers: The agent uses a system prompt generated via openuiChatLibrary.prompt to constrain the assistant's behavior to interactive UI generation.
- Capability inventory: The skill streams message chunks representing UI structures to a client-side FullScreen component for rendering.
- Sanitization: Content is encapsulated within HumanMessage objects before being passed to the LLM model, providing a structural layer of separation.
Audit Metadata