MuckScraper alternatives on desktop

XDA ran a piece this month on MuckScraper, a self-hosted news aggregator that groups articles across outlets, scores bias with a local LLM, and generates summaries against your own Ollama instance. It’s a genuinely clever project. It’s also a solo GitHub repo at grregis/MuckScraper with one contributor, no packaged releases, and a Docker Compose file that assumes a lot of local model wiring. Seven MuckScraper alternatives with real maintainer counts cover most of the same job without the Ollama dependency.

Quick comparison

App Best for Free plan Starting price Standout feature
FreshRSS The default self-hosted aggregator Yes Free Massive extension ecosystem
Miniflux Minimalist single-binary reader Yes Free or $15/yr hosted LLM integration since 2024
Selfoss Multi-source dashboard Yes Free Aggregates RSS, Twitter, Reddit
Tiny Tiny RSS Power-user filtering Yes Free Regex filters, plugin API
CommaFeed Google Reader clone Yes Free or $3/mo hosted Clean Java-based UI
Yarr Single-file desktop reader Yes Free Runs as a local binary
Feedbin Managed newsletter reader 14-day trial $5/mo E-books, newsletters, podcasts

Why people leave MuckScraper

The GitHub repo shows a single contributor and a handful of stars as of mid-July. Nothing wrong with a hobby project, but news aggregation is a job that runs 24/7, and choosing a solo repo for it carries risk. Three concrete pain points come up on the XDA piece’s comments and on Hacker News:

  1. Ollama dependency by default. MuckScraper’s bias scoring and summarization require a local model. If you don’t already run Ollama with a 7B model, you’re standing up two services, not one.
  2. No packaged releases. Docker Compose from source works if you know Docker, but there’s no Homebrew formula, no APT repo, no one-line installer. Every upgrade is a git pull.
  3. Story clustering is the only meaningful new idea. Cross-outlet grouping is nice, but every serious RSS reader in this list has been solving discovery and filtering for a decade. If bias scoring isn’t the killer feature for you, MuckScraper is a heavier version of tools you already had.

The alternatives

FreshRSS – Best default self-hosted news aggregator

FreshRSS is the community answer to Google Reader closing in 2013, and in 2026 it’s the default choice for anyone standing up a self-hosted feed reader for the first time. PHP-based, MySQL or PostgreSQL backend, Docker image officially maintained. Runs on any Raspberry Pi 3 or up.

The extension ecosystem is what sets it apart. Extensions for archived-article scraping, YouTube channel handling, XPath scraping for sites without RSS, and mobile companion apps (FeedMe, Readrops) cover almost every use case MuckScraper’s built-in features try to duplicate.

Where it falls short: No built-in LLM summarization. The default UI still looks like a 2015 web app. Setup requires a webserver in front.

Pricing:

Migrating from MuckScraper: Export your OPML from MuckScraper, import into FreshRSS in three clicks. Any bias scores you tagged in MuckScraper will not transfer.

Download: FreshRSS.org

Bottom line: Start here if you’re leaving MuckScraper and want the safest self-hosted default.

Miniflux – Best for minimalists who still want LLM integration

Miniflux is a single Go binary, a PostgreSQL backend, and nothing else. No plugins, no themes, no extensions. What it does have as of the 2.1 release cycle is native integration with third-party AI providers for on-demand article summarization. You point it at OpenAI, Anthropic, or a local Ollama endpoint, and it summarizes selected articles inline.

The reader UI is aggressive about readability. Full-text scraping is built in and enabled per feed. Keyboard shortcuts cover every action.

Where it falls short: No multi-user support in the self-hosted build (you get one account). No mobile app; the PWA works but isn’t native. The UI is spartan by design; some readers find it too plain.

Pricing:

Migrating from MuckScraper: OPML export imports cleanly. Configure your LLM provider in Miniflux settings to match the model you were using with MuckScraper.

Download: Miniflux.app

Bottom line: The best pick if you want the AI summarization MuckScraper offers without the Docker Compose gymnastics.

Selfoss – Best for aggregating beyond RSS

Selfoss treats every source as a “spout” and covers RSS, Twitter (via Nitter), Reddit, Mastodon, GitHub commits, and about a dozen other sources. It’s the closest tool on this list to MuckScraper’s cross-source ambitions, though without the LLM layer.

The dashboard is river-style: everything in one column, sorted newest-first. Filtering by tag or source is instant. The stack is PHP plus SQLite (or MySQL for larger installs), and it runs on almost anything.

Where it falls short: Development is slower now (releases every few months, not every few weeks). Twitter integration broke after API changes and only works through Nitter mirrors that are hit-or-miss. Mobile UI works but doesn’t sync notifications.

Pricing:

Migrating from MuckScraper: OPML import works for RSS. Non-RSS sources (Reddit, YouTube) need reconfiguring by hand.

Download: Selfoss.aditu.de

Bottom line: Pick Selfoss if you want RSS plus everything else in one dashboard.

Tiny Tiny RSS – Best for power-user filtering

Tiny Tiny RSS is the option for people who write regex filters on their news feed. Filter rules run on every article ingest and can auto-tag, auto-mark-read, or auto-star based on regex matches, date ranges, and source domains. The plugin API opens the door to anything the maintainer hasn’t built.

The project’s development moved to a self-hosted Gitea in 2020 (the developer feuded with GitHub) but is still active with regular releases.

Where it falls short: The install is PHP plus PostgreSQL and requires nginx tuning to run smoothly at scale. The default UI is dated. Some plugins are stale.

Pricing:

Migrating from MuckScraper: OPML import for RSS. Rebuild any tag rules using TT-RSS’s regex filter syntax.

Download: tt-rss.org

Bottom line: The power-user pick. Skip if you don’t already write regex.

CommaFeed – Best for the Google Reader look

CommaFeed is a Java-based self-hosted feed reader with a UI that visibly imitates Google Reader circa 2013. That’s the point. Three-pane layout, folder hierarchy, keyboard shortcuts that match the muscle memory of anyone who used Reader in the browser.

Runs as a single JAR or a Docker container. Uses PostgreSQL or MySQL for storage. Actively maintained by a single developer with a small contributor base.

Where it falls short: No advanced filtering. No plugin system. Java memory footprint is heavier than PHP or Go equivalents.

Pricing:

Migrating from MuckScraper: OPML import works. Any custom bias tags do not carry across.

Download: CommaFeed.com

Bottom line: Pick this if you miss Google Reader and want to run a self-hosted clone that looks like it.

Yarr – Best for a local-only single-binary reader

Yarr is the extreme minimalist option: one Go binary, one SQLite file, one browser tab. No server needed. You run yarr on your laptop, it opens localhost:7070 in your browser, and that’s your feed reader. Ideal for a single reader on a single machine without any hosting overhead.

The maintainer (Nazar Kanaev) ships builds for Windows, macOS, and Linux. Total install size is around 15 MB.

Where it falls short: No sync across devices. No multi-user. No LLM anything. Filtering is basic.

Pricing:

Migrating from MuckScraper: OPML import works. Any bias data does not.

Download: github.com/nkanaev/yarr

Bottom line: The right pick if you’re the only reader and you don’t want to run a server. Skip if you want cross-device sync.

Feedbin – Best managed alternative if you’ll pay

Feedbin is not self-hosted, and that’s the point. If MuckScraper feels like a hobby that turned into a chore, Feedbin is the paid opposite: managed hosting, a native iOS app, first-class newsletter routing (each account gets an @feedbin.me email that turns newsletters into feeds), and podcast subscription tracking.

Bias scoring isn’t there. Neither is LLM summarization. What Feedbin does have is uptime and a full-time small team maintaining it since 2013.

Where it falls short: Costs $5 a month. Not open source. Not self-hosted.

Pricing:

Migrating from MuckScraper: OPML import in one click.

Download: Feedbin.com

Bottom line: Pay Feedbin if you value your time more than $5 a month. Skip if self-hosted is a hard requirement.

How to choose

Pick FreshRSS if you want the safest self-hosted default. It has the largest community and the most extensions.

Pick Miniflux if you specifically want the LLM summarization angle that drew you to MuckScraper, without the Docker Compose complexity.

Pick Selfoss if aggregating RSS plus Reddit plus GitHub commits into one dashboard is more valuable than the news bias scoring.

Pick Tiny Tiny RSS if you write regex filters and want power-user control.

Pick Yarr if you want a single-user local-only reader and refuse to run a server.

Pay for Feedbin if you want to stop being your own sysadmin. It’s the honest answer for people who admit they’d rather read than maintain.

Stay on MuckScraper if the LLM bias scoring is the killer feature. No other tool on this list matches it out of the box.

FAQ

What is MuckScraper?

MuckScraper is a self-hosted news aggregator by grregis on GitHub. It groups articles across outlets into stories, scores bias with a local LLM, and generates summaries. It requires Ollama or a compatible local model server to run.

Is FreshRSS a good MuckScraper alternative?

Yes, for most people. FreshRSS covers the RSS-aggregator role better than MuckScraper does, has more contributors, packaged Docker images, and an ecosystem of extensions. It doesn’t cover the LLM bias scoring, which is MuckScraper’s differentiator.

Which self-hosted feed reader supports AI summarization?

Miniflux has built-in AI summarization since the 2.1 cycle. Point it at OpenAI, Anthropic, or a local Ollama endpoint and it summarizes articles inline. That’s the closest match to MuckScraper’s LLM feature in a maintained project.

Can I run any of these without a server?

Yes. Yarr is a single binary you run on your laptop that opens a browser tab as its UI. No server, no port forwarding, no Docker. It’s single-user only.

What runs on a Raspberry Pi?

FreshRSS, Miniflux, Selfoss, and Yarr all run comfortably on a Raspberry Pi 3 or newer. Tiny Tiny RSS runs but wants a Pi 4 for smooth performance. Feedbin is not self-hosted.