Overview
A Frontend SDK that enables developers to easily embed AI agents into web applications with first-class UX support. This toolkit provides plug-and-play components for chat, streaming actions, memory context, and tool invocation — ideal for SaaS, dashboards, and productivity tools.
Problem
Embedding agents in apps today is custom work: WebSockets, token streaming, prompt history, memory management — all handcrafted per use case. Developers need a standard toolkit to:
- Render chat UIs with token-by-token streaming
- Manage memory/session/state via local-first storage
- Call tools/functions safely from UI with feedback
Solution
The SDK includes:
<AgentChat />: Embeddable, themeable chat UI<MemoryStore />: Manages session memory + user feedback<ToolCall />: Wraps function calling UX with input forms<AgentConfigPanel />: Live parameter tuning for devs or end-users
Architecture
- Written in TypeScript + React
- Built-in support for OpenAI functions, Langchain tools, Supabase memory
- LocalStorage + Supabase dual persistence for agent memory
Use Cases
- Custom CRMs with embedded AI assistants
- Knowledge bases and helpdesks with streaming GPT support
- Personal dashboards (health, finances, habits) with interactive agents
Future
- Vue + Svelte versions
- PWA compatibility + offline fallback
- Plugins for voice input, camera input, and screen recording