Support & Documentation
Welcome to Local Agent! This extension brings AI-powered page analysis directly to your browser using Ollama, a local AI service running on your own machine. All processing happens locally - your data never leaves your computer.
📥 Installing Ollama & Downloading Models
Step 1: Install Ollama
1 Visit https://ollama.com
2 Download Ollama for your operating system (macOS, Windows, or Linux)
3 Run the installer and follow the installation prompts
4 Verify installation by opening Terminal (macOS/Linux) or Command Prompt (Windows) and typing:
ollama --version
Step 2: Download AI Models
Ollama needs AI models to function. Here are some recommended models:
Recommended for beginners:
- Phi-3 Mini (3.8B) - Fast, lightweight, great for most tasks
ollama pull phi3
- Llama 3.2 (3B) - Excellent balance of speed and quality
ollama pull llama3.2
For more powerful responses:
- Llama 3.1 (8B) - More capable, requires more RAM
ollama pull llama3.1:8b
- Mistral (7B) - Great for technical content
ollama pull mistral
⚠️ Note: Model downloads can be 2-5GB in size. Ensure you have sufficient disk space and a stable internet connection.
Step 3: Start Ollama with CORS Enabled
The extension requires Ollama to allow connections from Chrome. Start Ollama with CORS enabled:
macOS/Linux:
launchctl setenv OLLAMA_ORIGINS "*"
Windows (PowerShell):
setx OLLAMA_ORIGINS "*"
🚀 How to Use the Extension
Getting Started
1 Install the extension from the Chrome Web Store
2 Make sure Ollama is running with CORS enabled (see above)
3 Click the extension icon in your browser toolbar to open the side panel
4 Check the connection status - it should show "Ollama Connected" with a green indicator
Using @ Mentions to Reference Tabs
The extension's powerful feature is the ability to reference specific browser tabs in your questions:
- Type @ in the input field
- Autocomplete appears showing all your open tabs
- Select a tab using arrow keys (↑/↓) and press Tab or Enter
- Reference multiple tabs in one question
Example queries:
Summarize @"Research Article"Compare @"Product A" with @"Product B"How does @"Assignment Rubric" apply to @"My Essay"?
Keyboard Shortcuts
- Enter - Send your message
- Shift + Enter - New line in message
- ↑ / ↓ - Navigate autocomplete suggestions
- Tab or Enter - Select autocomplete suggestion
- Esc - Close autocomplete
Starting a New Chat
Click the "New Chat" button in the header to clear the conversation history and start fresh.
Changing Models
Use the Model dropdown at the bottom of the side panel to switch between installed Ollama models. Each model has different capabilities and speed.
🔧 Troubleshooting
Extension shows "Ollama Disconnected"
- Make sure Ollama is running
- Verify CORS is enabled (see Step 3 above)
- Click the "Reconnect" button (circular arrow icon)
- Check that Ollama is running on port 11434
"No models found" error
- Download at least one model using
ollama pull (see Step 2 above)
- Restart Ollama after downloading models
- Click "Reconnect" in the extension
Tab content not extracting
- Some pages (chrome://, extension pages) cannot be accessed for security reasons
- Try refreshing the tab and trying again
- Make sure you're using @ mentions to reference tabs
Slow responses
- Switch to a smaller model (phi3 or llama3.2:3b)
- Close other applications to free up RAM
- Consider using a GPU-accelerated model if you have a compatible GPU
🔒 Privacy & Security
Your data stays private:
- All AI processing happens locally on your machine
- No data is sent to external servers
- Content is only extracted from tabs you explicitly reference with @ mentions
- Conversation history is stored locally in your browser
📚 Additional Resources