Simple Ollama Chat

A lightweight, privacy-focused chat client for Ollama models, built with React and TypeScript. Use it to quickly experiment with different Ollama endpoints and models in a clean, responsive UI.

Features

• 💬 Chat with Ollama models using a clean, intuitive interface.
• 🔌 Connect to any Ollama API endpoint.
• 🔁 Switch models on the fly.
• 🧠 Toggle "thinking" mode for models that support it.
• ⚡ Streaming responses for a more dynamic, real-time experience.
• 🌗 Light and dark theme support.
• 🔒 Privacy-first, chat history is stored locally in the browser's localStorage.

Requirements

• Node.js (recommended via nvm) and npm or pnpm.

Quick Start

Run these commands to get started locally:

# Clone the repository
git clone https://github.com/jonigl/simple-ollama-chat.git
cd simple-ollama-chat

# Install dependencies
npm install

# Start the dev server
npm run dev

Open http://localhost:5173 (or the URL shown by Vite) in your browser.

Usage

• Select an Ollama model from the UI or enter a custom API endpoint.
• Send messages and receive streaming responses when supported by the model.
• Switch themes or toggle thinking mode via the settings panel.

This project was scaffolded using Lovable.

Privacy and Storage

All chat history is saved locally to the browser's localStorage by default. No data is sent to any third-party storage unless you explicitly configure a remote endpoint.

MCP Server Bridge

To effortlessly integrate and test Model Context Protocol (MCP) servers with this client, you can use the ollama-mcp-bridge: https://github.com/jonigl/ollama-mcp-bridge. The bridge acts as a transparent proxy in front of the standard Ollama API, enabling the seamless use of custom MCP servers alongside existing Ollama models. It introduces a dedicated endpoint that you can point this client toward, significantly simplifying development and testing of your MCP servers, whether they run locally or are accessed behind a network proxy.

You can point this client's Ollama endpoint directly at the bridge (for example, http://localhost:PORT). This configuration allows your client, the bridge, Ollama, and your MCP server to all run and interact locally. This creates a streamlined environment for end-to-end testing and development of your custom MCP server logic.

Tech Stack and knowledgements

• Vite
• React + TypeScript
• Tailwind CSS
• shadcn/ui components
• Lovable was used to create the initial project scaffolding.