# Chat Gun

A full-stack multi-agent chat application built with React, TypeScript and LangGraph.js, offering a deep researcher, a chat assistant, a math solver and an MCP tool-calling agent behind a gateway with streaming, image input and human-in-the-loop confirmation.

Chat Gun is a full-stack agent chat application built with React, TypeScript and LangGraph.js. It offers four agents: a Deep Researcher for multi-step research with citation checking, a Chat Assistant for general conversation, a Math Solver for calculations, and an MCP Agent that calls native and MCP tools.

It streams responses and tool activity in real time with cancellation and error handling, supports PNG, JPEG and WebP image input analysed by a vision model, and asks the user to confirm ambiguous choices, such as a weather location, before continuing. Native tools include a calculator and Tavily-powered web search, page fetching, and weather lookups, and it can optionally load Filesystem and Brave Search MCP servers.

The frontend runs on Vite and React, a backend-for-frontend service in Node.js handles API key authentication, CORS, request limits, timeouts and rate limiting, and the backend runs LangGraph.js agents against Qwen or any OpenAI-compatible or CCR-compatible model endpoint. It includes metrics, OpenTelemetry and optional Opik tracing for observability and evaluation.

Website: https://github.com/HsienW/chat-gun