Live demo · Create your reader
Your own git-backed snapshot of the feeds you follow.
Give aggr a website URL. It finds the feed, saves each new item as Markdown in your Git repository, and publishes a static site you can search and read on any device. The copy remains after the original page is gone. The published site is static files.
- Fork the working instance, keeping Copy the main branch only selected to start without the demo's archive.
- Replace
aggr.tomlwith the starter and edit the title and sources. A website URL is enough; aggr discovers the feed. - In Settings → Pages → Build and deployment, set Source to GitHub Actions.
- Open Actions, enable workflows, and run aggr on
main. The site ishttps://<you>.github.io/<repo>/until you set a domain.
Edit the configuration before the first run. The starter follows two sites and keeps each feed to 250 items, 730 days, and 1 GB, using the feed text and leaving images at the publisher. Original-page extraction and local media are settings you turn on when you want them. The default GitHub setup publishes the reader and its archive publicly. See sources, hosting, and build budgets.
To try it on your machine:
mise use -g github:aymericbeaumet/aggr # or download a binary from Releases
mkdir reads && cd reads && git init
aggr init
aggr dev # http://127.0.0.1:7319dev does not commit or push.
| Project | Focus |
|---|---|
| Miniflux / FreshRSS | A server that reads feeds for you. aggr stores the articles as files in Git and serves a static site. |
| Bubo | One generated page of feed links. aggr also keeps the article text. |
| wallabag | Pages you save, to read later. aggr follows sources on a schedule. |
| ArchiveBox | Broad snapshots of the web, including feeds. aggr is a reader whose history is Markdown in Git. |
Configuration stays on your primary branch. Articles live on a separate aggr branch: one
Markdown file per item, with the original URL and the time it was first captured. A sync that
changes nothing creates no commit, and one broken source does not stop the others. Items copied
from another aggr repository keep their original links. See
copying another instance.
website URL → scheduled sync → Markdown on the aggr branch → static reader
An open feed picks up a finished deployment on its own. Reading covers search, the phone layout, keyboard shortcuts, and offline copies. A stored item and the rules for editing the branch are in the Git model. How large the public demo is, and how long a run takes, is in benchmarks.
- Reading state stays in the browser that recorded it.
- A capture keeps the article or the feed entry. Feed text can be a summary, and original-page extraction can fail. Video and audio still come from the publisher.
- A public site republishes what it captured. Pages ask search engines to skip them unless you opt in, and that request leaves a public repository public. The original link stays on every item. See publication.
- Retention limits the current tree. Older commits remain for as long as that history is kept. See the Git model.
Sources · Reader · Commands · Hosting · Rendering · Git model · Interoperability · Performance
Rust generates the site. The reader is a Svelte application compiled ahead of time and shipped inside the binary, so building a site never needs Node. See contributing and client development.
MIT, for every version published. See licensing for contribution terms and third-party notices.