The complete article
RSSnook fetches and stores full text with each story, so truncated feeds become a calm reading view instead of a trail of tabs.
Software by Adrien Le Falher A quieter corner
rssnook
A quieter corner
of your internet.
A private, self-hosted RSS reader that brings the full article, your own organization, and local AI digests into one calm reading space. No cloud account, tracking, or hidden ranking.
- Free & MIT licensed
- Single-user by design
- Docker Compose
01 One inbox for the open web, ordered on your terms.
The reader, reclaimed
Your feeds are a library.
Not a slot machine.
RSSnook is built for choosing sources deliberately, finishing what is worth reading, and letting the rest go. It replaces the endless refresh with a library you can actually organize.
Your own structure
Put one feed in several topics, save searches, add colour-coded tags, and keep bookmarks separate from the read-later queue.
No ranking algorithm
Stories arrive from sources you chose. Browse by topic, feed, date, or search without a platform quietly rewriting the order.
One reading surface
See the whole signal.
Move from a broad daily scan to one topic, one saved search, or one full article without leaving the app.
Topics can overlap
A design feed can live in both “Art & Culture” and “Creativity & Design.” Organize by how you think, not by a rigid folder tree.
Reddit stays a conversation
Open a Reddit post with its live threaded comments, scores, and context instead of flattening it into a headline.
Keep what matters
Bookmark a reference, send something to read later, or highlight a passage and attach a private note.
Local intelligence
AI that reads with you.
Not about you.
Ollama runs the language model on your machine. Your subscriptions, reading history, and prompts do not need to become somebody else's training data to make the pile manageable.
A daily digest
At a time you choose, Ollama summarizes the last 24 hours topic by topic.
Summaries on demand
Ask for the short version of one article and keep the result cached locally.
Names that keep surfacing
Extract people, organizations, places, and subjects to see what is trending across your feeds.
Daily digest
Sunday · 19 July
Your morning, compressed.
Artificial intelligence12 stories
Open-source model releases moved toward smaller edge deployments, while new training work focused on longer contexts and reusable robotics datasets.
Photography6 stories
Two new camera firmware releases and a practical field review are worth a closer look.
Generated locally Ollama · Qwen
Less maintenance, less noise
Teach the inbox
your house rules.
Rules run as articles arrive. They do the repetitive sorting without deciding what you are allowed to see.
Automation / New ruleActive
WHEN
TopicisPhotography
Titlecontainsfirmware
THEN
★ Bookmark# Cameras↗ Priority +1
Automation rules
Auto-tag, bookmark, or mute new articles when a feed or title matches your rule.
Smart searches
Save a full-text query and let matching articles collect themselves.
Mute filters
Drop recurring noise by plain text or regular expression, globally or one feed at a time.
Signals you control
Use thumbs up and down to tune a personal priority score without exporting that feedback.
Private by location
Your reading history
has an address: home.
The application, database, extracted articles, and model all run in your Docker stack. There is no RSSnook account and no remote storage service to keep in sync.
SQLite databaseFull article archiveHighlights & notesOllama output
Built for a private network. RSSnook intentionally has no login screen. Keep it on your LAN or behind Tailscale; do not expose its port to the public internet.
Open, ordinary pieces
A stack you can inspect.
MIT-licensed source, plain Docker Compose, and a single SQLite file for your data.
InterfaceSvelteKit
→ APIFastAPI
→ ArchiveSQLite + FTS5
→ Local AIOllama
Good to know
Before you make a nook.
RSSnook is opinionated, self-hosted software for one person—not a hosted feed service.
01What is RSSnook?
RSSnook is a free, open-source, self-hosted RSS and Atom reader for one person. It fetches full articles, organizes feeds into topics, includes Reddit comments, and can create private AI summaries with a local Ollama model.
02Does RSSnook send my reading history to the cloud?
No. Articles, reading state, bookmarks, tags, searches, highlights, notes, and AI output are stored in your local SQLite database. Local summaries run through Ollama on your own machine.
03Can it follow sites that do not expose a normal RSS feed?
Often. The included RSSHub service can turn Bluesky, YouTube, and hundreds of other supported sources into feeds. Standard RSS and Atom URLs work directly.
04Can I import feeds from another RSS reader?
Yes. RSSnook imports and exports OPML. Folder names in an imported OPML file become topics, and bulk URL import can use the local language model to propose topic assignments for review.
05Is RSSnook safe to expose on the public internet?
No. RSSnook is a single-user application with no authentication layer. Run it on a private network or behind a private access tool such as Tailscale, and do not expose its port directly to the public internet.
06What do I need to run it?
Docker Compose is the simplest setup. The stack includes the SvelteKit interface, FastAPI backend, SQLite database, nginx, RSSHub, and Ollama. Ollama can run on CPU, or use a supported NVIDIA GPU when available.
Free · open source · self-hosted
Make room for what
you chose to follow.
Clone the project, start the Docker stack, and move your subscriptions over with OPML.