A universal remote for LangGraph agents

What it does

Agent Chat UI is a Next.js application that drops a chat interface on top of any LangGraph server exposing a messages key. Point it at a local dev server or a production deployment, pick your graph, and start talking to your agent. There’s a hosted version if you don’t feel like running anything locally.

The interesting bit

The project treats your agent like a black box with a streaming message contract. It doesn’t care what’s inside the graph—tool calls, multi-step reasoning, human-in-the-loop breakpoints—so long as the server speaks LangGraph’s protocol. The artifact rendering is the clever part: agents can emit structured content that opens in a side panel, turning a text stream into something closer to an interactive workspace.

Key highlights

• Connects to both Python and TypeScript LangGraph servers interchangeably
• Supports message visibility controls: suppress live streaming or hide messages permanently via ID prefixes and tags
• Renders agent-generated artifacts in a dedicated side panel using a React context hook
• Production path via an API passthrough package that proxies and authenticates requests server-side
• Can be bootstrapped with npx create-agent-chat-app or used directly at the deployed Vercel URL

Caveats

• Production requires reworking authentication; the default setup exposes LangSmith API keys to the client, which the README explicitly warns against
• The artifact rendering API is powerful but underspecified—implementing it means writing custom React components against a loosely documented hook contract

Verdict

Worth a look if you’ve already got LangGraph agents running and want a quick, standardized chat frontend without building one yourself. Skip it if you need deep customization of the UI or aren’t bought into the LangChain ecosystem.