samuelclay/newsblur

NewsBlur is a personal news reader that lets you subscribe to websites and blogs through their RSS feeds, which are a standardized way sites publish new content so readers can follow them without visiting each site manually. It shows you stories in a clean interface and can display the original web page alongside the article, giving you the full context instead of just a text summary. One of its more distinctive features is a training system. You can tell NewsBlur whether you like or dislike specific authors, topics, or keywords, and it will learn from those signals to automatically highlight stories you are likely to enjoy and de-emphasize ones you are not. It also has social features: you can share articles on a public page called a Blurblog and see what other users are sharing. NewsBlur is available as a web app at newsblur.com and has free native apps for iOS and Android. The free account tier supports up to 64 feed subscriptions, with paid plans for more. You can also run your own copy of NewsBlur using this repository and Docker, which means setting up and running the full application on a server you control rather than using the hosted service. The self-hosted setup is designed to start with just a few commands. The technology behind it is a Python web framework called Django for the backend, with several databases running in parallel: PostgreSQL for account data, MongoDB for storing articles, and Redis for caching. It also includes an MCP server, which allows AI assistants to connect to your NewsBlur account and interact with your feeds and stories on your behalf. The project has been running as a live service since at least the early 2010s and the full source code has been open from the start, which is uncommon for a consumer product that also operates as a paid subscription service.