Chrome extension that personalizes web pages based on your preferences, interests, and browsing history.

1. Extension captures page content and sends to local server via WebSocket
2. Server searches mem0 (Qdrant vector DB) for relevant user memories
3. Gemini analyzes content against memories, returns personalized modifications
4. Extension applies highlights, warnings, and tooltips to the page

Chrome Extension (content.js)
        ↕ WebSocket
Python Server (server.py)
        ↓
   ┌────┴────┐
   ↓         ↓
 mem0     Gemini
(Qdrant)  (analysis)

Stack: Python, FastMCP, mem0, Qdrant (local), Gemini, Chrome Extension (Manifest V3)

uv sync
cp .env.example .env
# Edit .env with your API keys
cd server && uv run python server.py

Environment variables (.env):

GEMINI_API_KEY=xxx          # Required
GOOGLE_MAPS_API_KEY=xxx     # Optional, for place enrichment

Load extension/ as unpacked extension in Chrome.

Connect to mem0 from MCP-compatible clients to seed preferences, add memories, or query what the system knows about you:

{
  "mcpServers": {
    "mem0": {
      "command": "/path/to/uv",
      "args": ["run", "--directory", "/path/to/humanize", "python", "mcp_server.py"]
    }
  }
}

Tools: add_memory_tool, search_memories, list_all_memories, delete_memory