🧠 Ollama Client — Chat with Local LLMs in Your Browser

Ollama Client is a powerful, privacy-first Chrome extension that lets you chat with locally hosted LLMs using Ollama — no cloud, no tracking. It’s lightweight, open source, and designed for fast, offline-friendly AI conversations.

🚀 Get Started — Install Now

❤️ Upvote Us on Product Hunt!

🌐 Explore More

✨ Features

🤖 Model Management

• 🔌 Local Ollama Integration – Connect to a local Ollama server (no API keys required)
• 🌐 LAN/Local Network Support – Connect to Ollama servers on your local network using IP addresses (e.g., http://192.168.x.x:11434)
• 🔄 Model Switcher – Switch between models in real time with a beautiful UI
• 🔍 Model Search & Pull – Search and pull models directly from Ollama.com in the UI (with progress indicator)
• 🗑️ Model Deletion – Clean up unused models with confirmation dialogs
• 🧳 Load/Unload Models – Manage Ollama memory footprint efficiently
• 📦 Model Version Display – View and compare model versions easily
• 🎛️ Advanced Parameter Tuning – Per-model configuration: temperature, top_k, top_p, repeat penalty, stop sequences, system prompts

💬 Chat & Conversations

• 💬 Beautiful Chat UI – Modern, polished interface built with Shadcn UI
• 🗂️ Multi-Chat Sessions – Create, manage, and switch between multiple chat sessions
• 📤 Export Chat Sessions – Export single or all chat sessions as PDF, JSON, Markdown, or Text
• 📥 Import Chat Sessions – Import single or multiple chat sessions from JSON files
• 📋 Copy & Regenerate – Quickly rerun or copy AI responses
• ⚡ Streaming Responses – Real-time streaming with typing indicators
• 🗑️ Message Deletion – Delete individual messages with cascading cleanup (v0.5.10)
• 📎 Export Individual Messages – Export single messages in PDF, Markdown, JSON, or Text formats (v0.5.10)
• 🌿 Conversation Branching – Fork conversations by editing messages and explore alternate paths

🧠 Embeddings & Semantic Search (Beta v0.3.0)

• 🔍 Semantic Chat Search – Search chat history by meaning, not just keywords
• 📊 Vector Database – IndexedDB-based vector storage with optimized cosine similarity
• 🎯 Smart Chunking – 3 strategies: fixed, semantic, hybrid (configurable)
• 🚀 Optimized Search – Pre-normalized vectors, caching, early termination
• 🔧 Configurable – Chunk size, overlap, similarity threshold, search limits
• 📁 Context-Aware – Search across all chats or within current session

📎 File Upload & Processing (Beta v0.3.0+)

• 📄 Text Files – Support for .txt, .md and text based files
• 📁 PDF Support – Extract and process text from PDF documents
• 📘 DOCX Support – Extract text from Word documents
• 📊 CSV Support – Parse CSV, TSV, PSV with custom delimiters and column extraction (Beta v0.5.0)
• 🌐 HTML Support – Convert HTML to Markdown for clean text extraction (Beta v0.5.0) with 50+ language support (Beta v0.5.0)
• ⚙️ Auto-Embedding – Automatic embedding generation for uploaded files
• 📊 Progress Tracking – Real-time progress indicators during processing
• 🎛️ Configurable Limits – User-defined max file size in settings

🌐 Webpage Integration

• 🧠 Enhanced Content Extraction – Advanced extraction with multiple scroll strategies (none, instant, gradual, smart)
• 🔄 Lazy Loading Support – Automatically waits for dynamic content to load
• 📄 Site-Specific Overrides – Configure extraction settings per domain (scroll strategies, delays, timeouts)
• 🎯 Defuddle Integration – Smart content extraction with Defuddle fallback
• 📖 Mozilla Readability – Fallback extraction using Mozilla Readability
• 🎬 YouTube Transcripts – Automated YouTube transcript extraction
• 📊 Extraction Metrics – View scroll steps, mutations detected, and content length

⚙️ Customization & Settings

• 🎨 Professional UI – Modern design system with glassmorphism effects, gradients, and smooth animations
• 🌓 Dark Mode – Beautiful dark theme with smooth transitions
• 📝 Prompt Templates – Create, manage, and use custom prompt templates (Ctrl+/)
• 🔊 Advanced Text-to-Speech – Searchable voice selector with adjustable speech rate & pitch
• 🌍 Internationalization (i18n) – Full multi-language support with 9 languages: English, Hindi, Spanish, French, German, Italian, Chinese (Simplified), Japanese, Russian
• 🎚️ Cross-Browser Compatibility – Works with Chrome, Brave, Edge, Opera, Firefox
• 🧪 Voice Testing – Test voices before using them

🔒 Privacy & Performance

• 🛡️ 100% Local and Private – All storage and inference happen on your device
• 🧯 Declarative Net Request (DNR) – Automatic CORS handling
• 💾 IndexedDB Storage – Efficient local storage for chat sessions
• ⚡ Performance Optimized – Lazy loading, debounced operations, optimized re-renders
• 🔄 State Management – Clean Zustand-based state management

🧩 Tech Stack

Built with React 18, TypeScript, Plasmo, Shadcn UI, Zustand, Dexie.js and Tailwind CSS.

For a deep dive into the code structure and design patterns, see ARCHITECTURE.md.

🛠️ Quick Setup

✅ 1. Install the Extension

👉 Chrome Web Store

✅ 2. Install Ollama on Your Machine

brew install ollama  # macOS
# or visit https://ollama.com for Windows/Linux installers

ollama serve         # starts at http://localhost:11434

💡 Quick Setup Script (Cross-platform):

For easier setup with LAN access and Firefox CORS support:

# Cross-platform bash script (macOS/Linux/Windows with Git Bash)
./tools/ollama-env.sh firefox   # Firefox with CORS + LAN access
./tools/ollama-env.sh chrome    # Chrome with LAN access

📄 Script file: tools/ollama-env.sh

This script automatically:

• Configures Ollama for LAN access (0.0.0.0)
• Sets up CORS for Firefox extensions (if needed)
• Shows your local IP address for network access
• Detects your OS (macOS, Linux, Windows) automatically
• Stops any running Ollama instances before starting

If you don't have the script file, you can download it directly or see the full setup guide: Ollama Setup Guide

More info: https://ollama.com

✅ 3. Pull a Model

ollama pull gemma3:1b

Other options: mistral, llama3:8b, codellama, etc.

⚙️ 4. Configure the Extension

• Click the Ollama Client icon

• Open ⚙️ Settings

• Set your:

  • Base URL: http://localhost:11434 (default) or your local network IP (e.g., http://192.168.1.100:11434)
  • Default model (e.g. gemma:2b)
  • Theme & appearance
  • Model parameters
  • Prompt templates

💡 Tip: You can use Ollama on a local network server by entering its IP address (e.g., http://192.168.x.x:11434) in the Base URL field. Make sure Ollama is configured with OLLAMA_HOST=0.0.0.0 for LAN access.

Advanced parameters like system prompts and stop sequences are available per model.

🛠️ Local Development Setup

Want to contribute or customize? You can run and modify the Ollama Client extension locally using Plasmo.

⚙️ Prerequisites

• Node.js (v18 or newer recommended)
• pnpm (recommended) or npm
• Ollama installed locally

📦 1. Clone the Repo

git clone https://github.com/Shishir435/ollama-client.git
cd ollama-client

📥 2. Install Dependencies

Using pnpm (recommended):

pnpm install

Or with npm:

npm install

🧪 3. Run the Extension (Dev Mode)

Start development mode with hot reload:

pnpm dev

Or with npm:

npm run dev

This launches the Plasmo dev server and gives instructions for loading the unpacked extension in Chrome:

• Open chrome://extensions
• Enable Developer mode
• Click Load unpacked
• Select the dist/ folder generated by Plasmo

🛠 4. Build for Production

pnpm build

⛓️ 5. Package for Production

pnpm package

🧪 6. Run, build and package in Firefox (Experimental)

Setup Ollama for Firefox:

Firefox requires manual CORS configuration. Use the helper script:

# Cross-platform bash script (macOS/Linux/Windows with Git Bash)
./tools/ollama-env.sh firefox

This configures OLLAMA_ORIGINS for Firefox extension support.

Build and run:

pnpm dev --target=firefox

pnpm build --target=firefox

pnpm package --target=firefox

Or with npm:

npm run dev -- --target=firefox

Load as a temporary extension.

📁 Code Structure

src/
├── background/        # Background service worker & API handlers
├── sidepanel/         # Main chat UI
├── options/           # Settings page
├── features/          # Feature modules
│   ├── chat/          # Chat components, hooks, semantic search
│   ├── model/         # Model management & settings
│   ├── sessions/      # Chat session management
│   ├── prompt/        # Prompt templates
│   └── tabs/          # Browser tab integration
├── lib/               # Shared utilities
│   └── embeddings/    # Vector embeddings & semantic search
├── components/        # Shared UI components (Shadcn)
└── hooks/             # Shared React hooks

Architecture: Feature-based organization with separation of concerns (components, hooks, stores). Zustand for global state, React hooks for local state.

✅ Tips

• Change manifest settings in package.json
• PRs welcome! Check issues for open tasks

💡 Recommended Models by Device

📦 Prefer quantized models (q4_0, q5_1, etc.) for better performance.

Explore: Ollama Model Library

🧪 Firefox Support

Ollama Client is a Chrome Manifest V3 extension. To use in Firefox:

1. Go to about:debugging
2. Click "Load Temporary Add-on"
3. Select the manifest.json from the extension folder
4. Manually allow CORS access (see setup guide)

🐛 Known Issues

• [ ] "Stop Pull" during model downloads may glitch
• [ ] Large chat histories in IndexedDB can impact performance

🗺️ Roadmap

See ROADMAP.md for planned features and upcoming releases.

🔗 Useful Links

• 🌐 Install Extension: Chrome Web Store
• 📘 Docs & Landing Page: ollama-client
• 🐙 GitHub Repo: Github Repo
• 📖 Setup Guide: Ollama Setup Guide
• 🔒 Privacy Policy Privacy Policy
• 🐞 Issue Tracker: Report a Bug
• 🙋‍♂️ Portfolio: shishirchaurasiya.in
• 💡 Feature Requests: Email Me

📚 For Developers & Contributors

• 🏗️ ARCHITECTURE.md - Technical architecture, design patterns, and implementation details
• 🗺️ ROADMAP.md - Planned features and development timeline
• 📊 PROJECT_ANALYSIS.md - Comprehensive code quality analysis
• 🤝 CONTRIBUTING.md - Guidelines for contributing to the project

📢 Spread the Word!

If you find Ollama Client helpful, please consider:

• ⭐ Starring the repo
• 📝 Leaving a review on the Chrome Web Store
• 💬 Sharing on socials (tag #OllamaClient)

Built with ❤️ by @Shishir435