AI-powered website aggregator.
Frontpage uses LLMs to scrape front pages, extract articles, and display them in a customizable dashboard. Add any website you want to follow, organize sources into columns, and get a single view of what's new across the web.
Features
- Customizable widgets
- Multiple dashboards
- Data stored on your machine as json files
- Supported providers: OpenAI (default), OpenRouter, Anthropic
- Quick start with docker-compose
- Open source and self-hosted
Motivation
Modern LLMs have become powerful enough and, more importantly, cheap enough to digest a website's front page and extract a list of articles — like an RSS feed, which many websites no longer provide. This project aggregates those results into a nice dashboard.
Running with Docker
docker run -d \
-p 3043:3043 \
-v ~/.frontpage:/data/frontpage \
-e OPENAI_API_KEY=your-key \
-e FRONTPAGE_MODEL=openai/gpt-5.4-nano \
hiquest/frontpage:latestOr with docker compose — edit docker-compose.yml, to set your API key (e.g., OPENAI_API_KEY) then:
The app will be available at http://localhost:3043. By default, dashboard configs are stored in ~/.frontpage, but you can override that by changing FRONTPAGE_HOME environment variable.
FAQ
Why not just use RSS?
Many websites have stopped providing RSS feeds, because, khmm, ads. Other times, RSS feeds are available but not frequently updated.
What model should I use?
The FRONTPAGE_MODEL value uses the format provider/model. Supported providers are openai, openrouter, and anthropic. For example:
openai/gpt-5.4-nanoopenrouter/google/gemini-3-flash-previewanthropic/claude-sonnet-4-6
The default is openai/gpt-5.4-nano. It seems to be good enough, and cheap enough. Make sure you set the corresponding API key environment variable (OPENAI_API_KEY, OPENROUTER_API_KEY, or ANTHROPIC_API_KEY).
So is this another service wrapped around a prompt?
Yes. You can read the prompt here.

