Self-host PriceSentinel on your own machine

Why self-host?

• No limits on products, store links, check intervals, AI jobs or channels.
• Your prices stay in a SQLite file on your own hardware.
• Email alerts through your own SMTP server, plus Gotify.
• A built-in MCP server so AI assistants can read prices and alerts.
• Local AI with Ollama or LM Studio, and private web search with SearXNG.
• Optional headless Chromium, domain rules and an HTTP proxy for difficult shops.

Desktop app for Windows, macOS and Linux

The desktop app runs the whole tracker on your computer, keeps checking prices from the tray when the window is closed and shows native notifications for alerts. No password is needed on your own computer.

Download the latest release on GitHub

The builds are not code-signed yet, so Windows SmartScreen and macOS Gatekeeper ask for confirmation the first time. Details in the desktop app documentation.

Run it with Docker Compose

git clone https://github.com/AndrasBogdanAndrei/pricesentinel.git
cd pricesentinel
docker compose up -d

Open http://localhost:8787 and choose an admin password. Data lives in the pricesentinel-data volume and the image has a health check. To add private web search for AI research, start the optional profile with docker compose --profile search up -d and set the SearXNG URL in Settings → AI.

If the instance is reachable from the internet, put it behind HTTPS and set SECURE_COOKIES=true. It fetches any URL you give it, so don’t share it with people you don’t trust.

Run it from source with Node.js

Requires Node.js 22.5 or newer (24 recommended).

git clone https://github.com/AndrasBogdanAndrei/pricesentinel.git
cd pricesentinel
npm install
npm run build
npm start

Configuration lives mostly in the app’s Settings; environment variables such as PORT, PRICESENTINEL_DATA_DIR and TZ are listed in the README configuration section.

Connect your AI assistant over MCP

A self-hosted PriceSentinel speaks the Model Context Protocol, so an MCP-capable AI assistant can read your prices, receive alerts and manage products. Create an API token in Settings → API & MCP and add the endpoint to your MCP client:

{
  "mcpServers": {
    "pricesentinel": {
      "type": "http",
      "url": "http://localhost:8787/mcp",
      "headers": { "Authorization": "Bearer ps_your_token" }
    }
  }
}

The server offers 17 tools (list products, price history, find deals, add products and links, check prices, alert inbox, AI research and more), resources such as pricesentinel://alerts/unread that push new alerts to connected clients, and prompts for a deal report or a weekly summary. A stdio bridge is included for clients that only launch commands. Full reference: MCP documentation.

MCP is not available on the hosted service for now.

Moving from the hosted service

Export your data from the settings page of your hosted account and import it into your own instance. Your hosted account keeps working until you delete it.