User Guide
This guide will help you get started with BrowserBee and show you how to use it effectively for your daily tasks.
#
InstallationClone the repository
git clone https://github.com/parsaghaffari/browserbee.gitcd browserbee
Install dependencies
npm install
Build the extension
npm run build
Load the extension in Chrome:
- Go to
chrome://extensions/
- Enable "Developer mode" in the top-right corner
- Click "Load unpacked" and select the
dist
directory - The BrowserBee options page should pop up automatically
- Go to
#
Setting Up Your LLM ProviderBrowserBee supports multiple LLM providers. You'll need to set up at least one to use the extension:
#
Anthropic Claude- Get an API key from Anthropic
- In the BrowserBee options page, select "Anthropic" as your provider
- Enter your API key in the field provided
- Select your preferred Claude model (e.g., Claude 3 Opus, Claude 3 Sonnet)
#
OpenAI- Get an API key from OpenAI
- In the BrowserBee options page, select "OpenAI" as your provider
- Enter your API key in the field provided
- Select your preferred model (e.g., GPT-4o)
#
Google Gemini- Get an API key from Google AI Studio
- In the BrowserBee options page, select "Gemini" as your provider
- Enter your API key in the field provided
- Select your preferred Gemini model
#
Ollama (Local Models)- Install Ollama on your computer
- Start the Ollama server with CORS enabled:
OLLAMA_ORIGINS=chrome-extension://YOUR_EXTENSION_ID ollama serve
- In the BrowserBee options page, select "Ollama" as your provider
- Enter the Ollama server URL (usually
http://localhost:11434
) - Add at least one custom model by specifying:
- Model ID (e.g.,
llama3.1
,qwen3
) - Display name (e.g.,
Llama 3.1
,Qwen 3
) - Context window size (default: 32768)
- Model ID (e.g.,
#
Using BrowserBee#
Opening the Side Panel- Click the BrowserBee icon in your Chrome toolbar
- The side panel will open on the right side of your browser window
#
Your First Command- Type a natural language instruction in the input field at the bottom of the side panel
- Press Enter or click the Send button
- Watch as BrowserBee carries out your instruction
Example commands:
- "Go to Google and search for 'browser automation tools'"
- "Find the latest news about artificial intelligence"
- "Check my Gmail inbox and summarize the unread messages"
- "Go to Twitter and show me the trending topics"
#
Understanding the InterfaceThe BrowserBee side panel consists of several components:
- Input Field: Where you type your instructions
- Message History: Shows the conversation between you and BrowserBee
- Token Usage: Displays the number of tokens used and the estimated cost
- Tab Status: Shows which tab BrowserBee is currently controlling
#
Approving ActionsFor certain sensitive actions (like making purchases or posting on social media), BrowserBee will ask for your explicit approval:
- BrowserBee will display an approval dialog
- Review the action it wants to take
- Click "Approve" to allow the action or "Deny" to prevent it
#
Common Tasks#
Web ResearchBrowserBee excels at gathering information from multiple sources:
Go to Google and search for recent developments in quantum computing. Visit the top 3 results and summarize the key points from each site.
#
Social Media ManagementBrowserBee can help you manage your social media presence:
Go to Twitter, check my notifications, and summarize any mentions or replies I've received in the past 24 hours.
#
Online ShoppingBrowserBee can assist with finding products and comparing prices:
Search for wireless noise-canceling headphones on Amazon, find the top 5 options under $200, and compare their features and ratings.
#
Email ManagementBrowserBee can help you manage your email:
Go to Gmail, find all unread emails from the last week, and create a summary of each one.
#
Tips and Tricks#
Be SpecificThe more specific your instructions, the better BrowserBee can help you:
- Instead of: "Find flights"
- Try: "Go to Kayak and find flights from New York to London for next weekend, prioritizing morning departures"
#
Use Step-by-Step InstructionsFor complex tasks, break them down into steps:
First, go to my Google Drive. Then, find the folder named "Work Projects". Next, create a new Google Doc titled "Project Plan". Finally, type "# Project Plan" as a heading.
#
Leverage MemoryBrowserBee remembers previous interactions with websites, making repeat tasks more efficient:
Go to my company's HR portal and submit my timesheet for this week, just like I did last week.
#
Troubleshooting#
BrowserBee Can't Access a Website- Make sure you're logged in to the website if it requires authentication
- Check if the website has anti-bot measures that might be blocking BrowserBee
- Try refreshing the page and trying again
#
Commands Not Working as Expected- Be more specific in your instructions
- Break complex tasks into smaller steps
- Check if the website's layout has changed since your last visit
#
High Token Usage- Use more concise instructions
- For repetitive tasks, create memories to reduce token usage
- Consider switching to a more cost-effective LLM provider
#
Privacy and SecurityBrowserBee is designed with privacy in mind:
- Your data stays in your browser (except for what's sent to the LLM provider)
- API keys are stored locally in your browser
- BrowserBee can't access your browser history or other sensitive information unless you explicitly instruct it to
However, be aware that:
- Text sent to LLM providers is subject to their privacy policies
- Be cautious when asking BrowserBee to interact with sensitive websites like banking portals