A Chrome extension that helps you browse with clarity: scrape and score article links, automate page actions with natural language, fact-check content (including videos), speak commands with voice, and optionally save to Gmail, Google Docs, and more.

• Toggle sidebar on any page with Ctrl+Shift+S (Windows) / Cmd+Shift+S (Mac)
• Light/dark mode – follows system preference
• Central hub for scraping, automation, fact-checking, and chat

• Article links – Fetches and scores links from the current page (Browserbase + optional Redis cache)
• Caching – Redis cache for faster repeat visits

• Natural language commands – Describe what you want in plain English:
  • "Click the login button"
  • "Hide the weather card"
  • "Remove all shorts from my YouTube feed"
  • "Remove ads from this page"
• Self-evolving agent – On script failure, the system retries with error feedback (Planner → Writer → Daytona validation)
• Safe Browsing Mode – Guardrails block truly destructive commands (e.g. "remove everything") while allowing targeted removals (remove shorts, ads, sidebars)

• Verify page, screenshot, or video – Content-based routing:
  • Video URLs (YouTube, Instagram, TikTok, etc.) → Reka Vision extracts claims from audio + visuals
  • Screenshot → Bem or Gemini image extraction
  • Text/Caption → Fastino or Gemini text extraction
• Fact Hunter – Searches trusted sources (Reuters, AP, BBC, etc.) via Tavily
• Truth Judge – Multi-model jury (Gemini + Reka) returns verdicts: True / False / Unverified
• Page trust score – 0–100 aggregate
• Deep Research – Yutori Research API for in-depth claim investigation (5–10 min); poll for results

• Watch pages – Create scouts to monitor URLs (e.g. "alert when article updated")
• Yutori Scouting API – Periodic checks, webhook notifications
• Email alerts – Optional Composio webhook for email when changes detected

• Speech-to-text – Microphone button sends audio to backend
• Transcription – MiniMax (preferred) or OpenAI Whisper
• Safety check – Fastino classifies voice input before execution

• Save to external apps – When you explicitly ask:
  • Gmail – Draft or send emails
  • Google Docs – Create or append summaries
  • Google Sheets – Add rows
  • Google Calendar – Create events
  • Notion – Create pages
• Chat follow-up – In multi-turn chat, DOM commands (remove, hide) run automation; save/email requests run Composio

├── backend/                 # FastAPI app (port 8001)
│   ├── agents/
│   │   ├── automation/      # Planner, writer, checker, validator (ADK)
│   │   └── truth/           # Fact Hunter, Judge (Gemini + Reka)
│   ├── routes/              # automate, scrape, voice, verify, scout
│   └── services/            # Composio, Bem, Reka Vision, Fastino, Yutori, Daytona
├── Browserbase/             # Scraper API (port 8000) – optional
├── BrowserPlugin/           # Chrome extension (manifest v3)
│   ├── icons/               # 16, 48, 128
│   ├── background.js        # Service worker (API calls, voice upload)
│   ├── content.js           # Injects sidebar
│   └── sidebar.js           # UI logic, scrape, automate, verify, voice
├── .env.example             # Copy to .env and fill in keys
├── run_backend.bat          # Start backend on 8001
└── run_scraper.bat          # Start Browserbase scraper on 8000

• Python 3.10+ and a virtual environment
• Chrome (or Chromium-based browser)
• API keys (see Environment)

1. Clone and enter the repo

   git clone https://github.com/Anantaverma20/Conscious-Web.git
   cd Conscious-Web

2. Create a virtual environment and install dependencies

   python -m venv venv
   venv\Scripts\activate          # Windows
   # source venv/bin/activate      # macOS/Linux
   pip install -r backend/requirements.txt

   For the Browserbase scraper (port 8000):
   pip install -r Browserbase/requirements.txt

3. Environment

   • Copy .env.example to .env
   • Fill in at least:
     • BROWSERBASE_API_KEY, BROWSERBASE_PROJECT_ID (scraping)
     • GOOGLE_API_KEY (Gemini, for automation)
     • TAVILY_API_KEY (News Truth fact-check; required for Verify)
     • DAYTONA_API_KEY (automation script validation)
   • Optional: MINIMAX_API_KEY or OPENAI_API_KEY (voice), REKA_API_KEY (video fact-check + jury judge), YUTORI_API_KEY (scout + deep research), etc.

4. Start the backend

   • Backend (required): port 8001
     run_backend.bat or:
     uvicorn backend.main:app --host 127.0.0.1 --port 8001
   • Scraper API (optional): port 8000
     run_scraper.bat or:
     uvicorn Browserbase.api:app --host 127.0.0.1 --port 8000

5. Load the extension

   • Open Chrome → chrome://extensions
   • Enable Developer mode
   • Click Load unpacked and select the BrowserPlugin folder

• Toggle sidebar: Ctrl+Shift+S (Windows) / Cmd+Shift+S (Mac), or use the extension icon
• Article links: Open the sidebar; it scrapes the page and shows links (backend on 8001; scraper on 8000 if used)
• Automation: Type a command (e.g. "Remove all shorts", "Hide the sidebar") and click Run or press Enter
• Voice: Click the microphone, speak your command, click again to stop; then Run to execute
• Verify Truth: Click the shield icon to fact-check the current page or screenshot
• Deep Research: After verifying, use Deep Research for a claim to run an in-depth Yutori investigation (5–10 min)
• Scout: Create scouts from the sidebar to watch pages for changes

Full list and comments are in .env.example. Do not commit .env.

See repository for license information.