Sparkfeed Documentation

Sparkfeed brings every blog, news source, and website your team cares about into one clean, distraction-free space. No algorithms, no ads, no telemetry. It is open source under AGPL-3.0 and runs entirely on infrastructure you control.

Start here

New to Sparkfeed? These three pages take you from nothing installed to a running instance.

Introduction What Sparkfeed is, why it exists, and what you can do with it. Local setup Clone the repo, install dependencies, and have Sparkfeed running in minutes. Demo mode Explore a seeded instance immediately. No account, no migration, no keys.

Explore by topic

The rest of the documentation, grouped by what you are trying to do.

Core concepts How RSS, feed aggregation, and the local database fit together. Features The reading interface, offline access, favorites, and responsive behaviour. Spark AI Summaries, trend detection, and categorization that run on a key you own. Configuration AI providers and every environment variable Sparkfeed reads. Tech stack React, TanStack Router, Nitro, SQLite, Drizzle, and Zustand. Developer Connect an AI agent over MCP, and manage the API keys that authenticate it.

Key features

• Any site to feed

  Turn any website into a trackable feed, even sites that never published RSS.

• Privacy first

  Zero telemetry, no ads, and your data never trains AI models.

• Self-hostable

  Open source under AGPL-3.0. Run it on your own infrastructure.

• Local first

  Feeds and articles live in a local SQLite database, ready for offline reading.

• Folders and keyword filters

  Organize sources hierarchically and cut the noise.

• OPML import and export

  Bring your subscriptions in and out. You are never locked in.

• Spark AI

  Summaries and trend detection that run on an AI key you control.

• MCP for AI agents

  Let Claude or Cursor search your feeds and read full articles over the Model Context Protocol.

• Fast and minimal

  A React and Tailwind interface built for near-instant navigation.

Local first, by default

Everything Sparkfeed knows lives in one SQLite file on your own disk. There is no account to create, no sync service to trust, and nothing phoning home. Point it at a Turso database when you want it on a server, or leave it exactly where it is and read offline.

Terminal

$ bun dev

✓ database    file:rss.db
✓ migrations  up to date
✓ feeds       24 sources refreshed

➜ ready on http://localhost:3000

On the roadmap

• Team workspaces Soon

  Shared feeds, folders, and role-based access.

• Public folder sharing Soon

  Publish a curated folder as a mini-publication via a public link.

• Webhooks Soon

  Push new articles to Slack, automation platforms, or your own endpoint.

How Sparkfeed compares

How Sparkfeed lines up against popular self-hosted and hosted readers.

Feature	Sparkfeed	FreshRSS	Miniflux	NewsBlur	Feedly	Inoreader
Open source	AGPL-3.0	AGPL-3.0	Apache-2.0	MIT (core)	No	No
Self-hostable	Yes	Yes	Yes	Yes †	No	No
Zero telemetry	Yes	Yes	Yes	No	No	No
Any site to feed	Yes	No	No	No	No	Yes ‡
OPML import / export	Yes	Yes	Yes	Yes	Yes	Yes
Bring-your-own AI keys	Yes	No	No	No	No	No
Team workspaces	Yes ◊	No	No	No	Yes ‡	Yes ‡

† NewsBlur self-hosting is supported but more involved to set up.  ‡ Available on paid or higher tiers.  ◊ Sparkfeed team workspaces are on the roadmap (see above).

Want the full breakdown, including licenses and privacy posture? Read The Best Private RSS Readers in 2026 and Feedly Alternatives That Respect Your Privacy.

Start reading in five minutes.

Get started