AI NewsOpen sourceAnnouncement
Horizon builds a daily AI news brief from the feeds you choose
Horizon is an open-source tool that reads your news feeds, scores items with an AI model, and sends one brief.
Image: GitHub
Why it mattersA team that pays for a news digest service can run its own. The editorial rules live in a file, which decides what gets picked and how it is written.
Most people who follow software news read the same ten sources every morning and still miss the one that matters. Horizon is an open-source tool that reads them for you. It writes one brief instead.
The project is by GitHub user Thysrael, whose profile gives the name Sheng Qiang. The repository is Python, under the MIT licence. It ranked 24th on Trendshift's daily list on the day this item was written.
Horizon's own GitHub page shows 11,346 stars and 1,695 forks. GitHub's API returns the same star number, which is a total since the project started about nine months ago.
What it does
The README says Horizon pulls from RSS, Hacker News, Reddit, Telegram, X, GitHub and financial news. It scores items, filters them against your rules, and removes duplicates. The brief can be written as Markdown or published to a GitHub Pages site. It can also be sent to an email inbox, a Feishu or WeChat room, or any webhook.
The idea the project is built around is a profile: one set of editorial rules for one kind of content. The README lists three built-in profiles. tech-news asks for the event and the context. tech-blog asks for background, the solution and the practical takeaway. ai-creator asks for a summary, with timely hooks and content angles when useful. A source gets assigned a profile, or the model picks one for it.
How a team would use it
The README names a cap on brief length. There is a second cap on how much space each category takes. So one busy topic cannot push every other story out. Horizon supports Claude, GPT, Gemini, DeepSeek, Doubao, MiniMax and local Ollama models. An interactive wizard writes the first config from questions about your interests.
Horizon also runs as a Model Context Protocol server. A Claude or Codex session can ask it to produce a brief instead of running the CLI by hand. There is a separate agent skill that installs the same way.
Why now
The repository has been live since early this year, and the activity picked up recently. The commit log shows a Firecrawl content extractor landed in the last few days. There are also recent fixes for sampling parameters across model providers, and tighter checks on scrape responses. The project has no tagged releases on GitHub. So this reads from the commit log, not from a release note.
A reader who already builds AI news digests internally can swap in Horizon. The profile files are Markdown, which means the editorial rules can change without touching Python. A reader who pays for a commercial news digest service now has an open-source option to put next to it.
Reveneau's take: the thing worth copying here is the profile idea. Most "AI news digest" projects let you pick sources and nothing else, so every item comes out shaped the same. Horizon keeps the sources and the output rules in separate files. That is how a careful editor would build it by hand.
Source
- Thysrael/Horizon (GitHub repository, README and commit log)
- Live demo (project's own site)
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.
