Web Keyword Sensor is a Home Assistant app that checks web pages on a schedule and publishes the results as automatically discovered MQTT entities. It can look for an exact phrase or use an AI model to interpret page content and find relevant information.
- Multiple independent page checks.
- Exact, case-sensitive or case-insensitive phrase matching.
- AI context matching with OpenAI, Google Gemini, or Anthropic Claude.
- Binary sensors for yes/no results and text sensors for AI summaries.
- Weekday and daily time-window scheduling, including overnight windows.
- No-auth checks, username/password/TOTP login, and interactive Browser SSO.
- MQTT discovery, immediate per-check testing, and provider connection testing.
- Bounded concurrency, request sizes, browser sessions, and timeouts to protect Home Assistant from slow or unusually large pages.
In Home Assistant open:
Settings → Apps → App Store → ⋮ → Repositories
Add the GitHub repository URL. The app will then appear under Local apps or as a custom repository app.
Install and configure Web Keyword Sensor, then open its Ingress panel to
create checks and AI provider profiles. The complete feature and configuration
guide is in web-keyword-sensor/README.md.
† This app was developed almost entirely using Ai.