Eclipse is an intelligent browser extension that revolutionizes how you interact with web content. Built with a multi-agent AI system, it combines powerful content analysis with browser automation to transform your browsing experience from manual tasks into intelligent workflows.
100% Open Source • Privacy First • Free • BYOK (Bring Your Own Keys)
Eclipse can automate browser tasks like filling forms, navigating websites and completing tasks. It uses AI to understand web pages and perform actions just like a human would. In the demo, it's being used for lead extraction from a conference portal, a common task in B2B/enterprise GTM and sales teams - reducing tasks that typically take 15-30 minutes down to just 2-4 minutes.
Powered by dom-engine, our in-house library that turns website DOMs into actionable context for AI agents.
Organize your browser tabs efficiently. Eclipse can group related tabs, switch between tabs, and manage multiple open tabs. Let the AI organize them automatically or specify how you want them grouped, including choosing colors.
Search YouTube videos and get AI-powered analysis. Perfect when you remember a phrase or moment from a video but not the title. Eclipse finds the right videos and pinpoints exact timestamps.
Extract and analyze content from your open browser tabs. Eclipse can process single or multiple tabs simultaneously. Also useful for searching text fragments across tabs and documents. Limited to 5 tabs per prompt for optimal context management.
Get AI analysis of YouTube videos with summaries, timestamps, and key moments. Ask Eclipse to summarize, search phrases, generate quizzes, or translate content. Perfect for research and learning.
Eclipse operates in two distinct modes to match your needs:
Full-featured mode with complete browser automation capabilities. This mode includes everything from Ask Mode plus tab management and browser automation. It operates through a sophisticated multi-agent system:
- Planner Agent: Understands your query and creates an actionable plan
- Executor Agent: Executes the necessary tools to complete your goal
- Validator Agent: Keeps the executor on track, manages errors, and provides feedback to ensure successful task completion
The multi-agent system works collaboratively to break down complex tasks into manageable steps and ensure reliable execution, with a maximum of 12 iterations as a safety mechanism to prevent infinite loops.
graph TD
A[User sends message] --> B[Orchestrator receives request]
B --> C[Orchestrator calls Planner]
C --> D{Planner evaluates request}
D -->|Direct response| E[Return response to user]
D -->|Needs plan| F[Planner creates plan with steps]
F --> G[Orchestrator executes executePlanWithValidation]
G --> H[Start glow animation if needs browser control]
H --> I[Execution loop MAX_ITERATIONS=12]
I --> J[Executor receives current step]
J --> K{Executor decides action}
K -->|Tool call| L[Execute specific tool]
K -->|No tool| M[Pause plan]
L --> N[Record result in toolCallHistory]
N --> O[Validator evaluates result]
O --> P{Validator determines state}
P -->|Step completed| Q[Advance to next step]
P -->|Step in progress| R[Continue with same step]
P -->|Plan completed| S[Finish execution]
Q --> T{More steps?}
T -->|Yes| I
T -->|No| S
R --> I
S --> U[Stop glow animation]
U --> V[Return final response to user]
M --> W[Return pause response]
Key Features:
- Maximum 12 iterations to prevent infinite loops
- Glow animation for visual feedback during browser control
- Complete audit trail of all tool calls and actions
- Dynamic plan updates based on validator feedback
- Robust error handling and recovery mechanisms
Includes all features except those that require direct browser interface actions (tab management and browser automation). Think of it as Eclipse's safe mode - use this mode when you need to analyze content, search videos, extract information, or get insights from your web pages.
- Groq API Key: Required for AI features
- YouTube API Key: Required for video search and analysis features
- Clone this repository
- Run
npm installto install dependencies - Run
npm run build:extensionto build the extension - Open Chrome and go to
chrome://extensions/ - Enable "Developer mode"
- Click "Load unpacked" and select the
distfolder - The Eclipse extension will appear in your browser toolbar
Note: Chrome Web Store submission is currently in process, but you can install manually using the steps above.
- Tab Management: Group tabs, organize by topic, custom colors
- Video Search & Analysis: YouTube search, AI analysis, timestamp detection
- Content Extraction: Single/multiple tab content extraction with text fragment search
- Video Analysis: Summarization, phrase search, quiz generation, translation
- Browser Automation: Currently in experimental version
The next steps include deploying MCP servers on GCP and integrating them with the extension for enhanced functionality.
Contributions are welcome! Please feel free to submit issues, feature requests, or pull requests to help improve Eclipse.
Eclipse is an open-source project built with the goal of making web browsing more intelligent and efficient. It was made open source to keep the door open for innovative ideas and provide an accessible, cost-effective alternative without the privacy concerns of Perplexity Comet.