What problem does it solve?
Integrates the OpenAI ChatKit UI with custom backends and domain-specific AI agents so conversational requests carry authentication, user/page context, and conversation history for accurate agent responses.
Core Features & Use Cases
- Backend adapter: Extend ChatKitServer and implement respond() to load thread history, extract metadata, build a single prompt for agents, and stream agent responses.
- Frontend integration: Provide a custom fetch interceptor to add auth headers and metadata, detect and wait for ChatKit script loading, extract page context and support text-selection "Ask" flows in React/Docusaurus apps.
- Production readiness: Support conversation persistence, multi-tenant isolation via RequestContext, graceful degradation when services fail, and connection pool warmup to avoid long cold starts.
- Use Case: Embed ChatKit in an educational Docusaurus site so authenticated learners can ask questions about selected page content and have those conversations persisted per user.
Quick Start
Add a custom fetch wrapper that injects X-User-ID and pageContext into ChatKit requests and implement ChatKitServer.respond to assemble history and run your agent.