Feedlynx

Self-hosted link collecting, honestly reviewed. No marketing fluff, just what you get when you run a Rust binary on your server.

TL;DR

• What it is: A self-hosted Rust binary that turns links you save into an Atom/RSS feed — no new app to learn, no new UI to check [README].
• Who it’s for: RSS readers who already live in Miniflux, FreshRSS, or NetNewsWire and want their read-later queue to live in the same place [README][5].
• Cost savings: Pocket Premium was $4.99/mo; Readwise Reader runs $7.99/mo. Feedlynx runs on a cheap VPS you likely already have, for free [README].
• Key strength: The architecture is the feature — your saved links become RSS entries, which means your existing feed reader handles everything: sync, mobile, search, read/unread state [README].
• Key weakness: 107 GitHub stars, one maintainer, no web UI, a hard cap of 50 entries in the feed, and no persistence for items once you’ve marked them read in your RSS reader [README][1].

---

What is Feedlynx

Feedlynx is a small HTTP server written in Rust. You run it on a machine you control, point a browser extension or Apple Shortcut at it, and when you add a link, the server fetches the page, extracts a title and description, and appends a new entry to an Atom feed file on disk. Your RSS reader subscribes to that feed. That’s the entire product [README].

The author, Wesley Moore (wezm), describes it plainly in the README: “Collect links to read or watch later in your RSS reader.” There is no dashboard. There is no mobile app. There is no tagging system, no highlights, no annotations. You get an Atom file and an HTTP endpoint [README].

What makes this interesting rather than just minimal is the architecture decision. Every read-later tool from Pocket to Instapaper to Omnivore solves the problem the same way: build a new app with its own reading interface. Feedlynx makes the opposite bet — that people who care enough about read-later to self-host it probably already have an RSS reader they prefer, and the job of a read-later tool is to put items into it, not compete with it [README][5].

The project has been around since at least mid-2024 (listed in selfh.st’s new software directory in August 2024 [1]) and had 108 stars at time of review. It’s a genuine one-person project — Wesley Moore is a developer known in the Rust and BSD communities, and feedlynx is available as a crate on crates.io [README]. The Dockerfile and CI (Cirrus CI) suggest this is maintained thoughtfully rather than abandoned.

---

Why people choose it

The honest answer: the community is small enough that “why people choose it” comes from the design decisions, not from a body of published reviews. The sources available [1][2][5] are directory listings and link roundups, not user testimonials or head-to-head comparisons.

What the directory listings do confirm is the audience. selfh.st called it “Generate RSS feeds from collected links via browser extension” [1] and jutty.dev linked to it with the description “Collect links to read later into an RSS feed” [5] — both reaching the same people: RSS-first users who resent context-switching to a separate read-later app.

The case for Feedlynx over established alternatives usually comes down to three things.

Versus Pocket. Mozilla announced the shutdown of Pocket in mid-2025. Anyone who had years of saved links on Pocket needed a replacement. Feedlynx doesn’t replicate Pocket’s feature set — no permanent archive, no full-text search, no tagging. But it solves the core use case (save now, read later in your reader) without depending on a third-party service that can be shut down [README].

Versus Wallabag. Wallabag is the canonical self-hosted Pocket replacement. It has a full web reader, mobile apps, tags, full-text search, and permanent archival. It also requires PHP, a database, and considerably more infrastructure. Feedlynx is a single binary. If you want the Wallabag feature set, use Wallabag. If you want to save links and read them in the RSS reader you already use, Feedlynx is significantly simpler to operate [README].

Versus Omnivore. Omnivore was a well-regarded open-source read-later service that shut down in 2024 after failing to find a business model. Its shutdown is exactly the scenario Feedlynx insulates you from — a single-binary tool with a file on disk doesn’t have a business model to fail [README].

The philosophical through-line: Feedlynx exists for people who trust RSS infrastructure more than product companies.

---

Features

Core:

• Serves an Atom feed from a local file [README]
• Fetches page title and description on link submission (live fetch at save time) [README]
• YouTube links get a video embed generated rather than just a title [README]
• Two-token auth: one private token to authenticate submissions, one feed token embedded in the feed URL path [README]
• Feed trimming: retains up to 50 entries; entries older than 30 days are removed when count exceeds 50, oldest first [README]
• Token generation: feedlynx gen-token prints a cryptographically random token [README]
• Logging via FEEDLYNX_LOG env variable, standard Rust log levels [README]

Platform support:

• Pre-compiled binaries for FreeBSD 13+ amd64, Linux x86_64, Linux aarch64, macOS Universal, Windows x86_64 [README]
• Docker (Dockerfile included in repo) [README]
• Runs on BSD, Linux, macOS, Windows [README]

Integrations (client side):

• Browser extension: https://github.com/wezm/feedlynx-ext [README]
• Apple Shortcuts for iOS/iPadOS/macOS (system share sheet integration) [README]
• Android via HTTP Request Shortcuts app (open-source, configured manually per the wiki) [README]

What’s absent:

• No web UI for browsing or reading saved items
• No full-text archival (the Atom entry contains a description extracted from the page, not the full page content)
• No tagging, folders, or organization beyond the feed itself
• No sync between multiple Feedlynx instances
• No search within the feed (your RSS reader handles this)
• No persistence once an item ages out — if you save a link and don’t read it within the trimming window, it disappears from the feed

---

Pricing: SaaS vs self-hosted math

Feedlynx has no SaaS tier. There is no hosted version and no commercial offering. The comparison is Feedlynx self-hosted versus the paid read-later market.

Read-later SaaS pricing (for reference):

• Pocket Premium: was $4.99/mo before the service closed [data not available from provided sources — pricing from public knowledge]
• Readwise Reader: $7.99/mo (or $96/yr)
• Instapaper Premium: $2.99/mo
• Matter: free tier with limits

Feedlynx self-hosted:

• Software: $0 (Apache-2.0 / MIT dual license) [README]
• Infrastructure: $0 if you have a VPS or home server already; $4–6/mo for a small Hetzner or Contabo VPS if you don’t
• RAM requirement: minimal — a single Rust binary serving one user needs well under 100MB

If you’re currently paying $7.99/mo for Readwise Reader specifically for the read-later queue (not the email digests or highlights — just the save-for-later), the self-hosted math is obvious: $0 software on $0 incremental infrastructure if you already run a server. Over a year against Readwise Reader: roughly $96 saved.

The realistic caveat is that most people evaluating Feedlynx are comparing it to free tiers of Pocket/Instapaper or to tools they’ve already sideloaded into their RSS reader setup. The cost argument is real but secondary to the workflow and data-sovereignty argument.

---

Deployment reality check

This is one of the simpler self-hosted deployments in the category. The entire state is a single Atom XML file on disk. No database. No Redis. No background job queue [README].

Minimum requirements:

• Any Linux/macOS/BSD/Windows host with network access
• Two environment variables set: FEEDLYNX_PRIVATE_TOKEN and FEEDLYNX_FEED_TOKEN (both must be 32+ characters, generated with feedlynx gen-token)
• A path to write the feed file

With Docker: The repo includes a Dockerfile [README]. A minimal compose file pointing at a volume for the feed XML, with the two required env vars, is all you need.

With a reverse proxy: Feedlynx binds to 127.0.0.1:8001 by default. To expose it externally (required for the browser extension to reach it from other devices), you need nginx or Caddy in front of it with HTTPS. This is standard homelab territory — not a Feedlynx-specific concern — but it’s a step that non-technical users will need help with.

Realistic setup time:

• Existing homelab with a reverse proxy already configured: 15–30 minutes including Docker setup and browser extension install
• Fresh VPS with no existing infrastructure: 1–2 hours including nginx/Caddy setup and domain pointing
• Non-technical user with no Linux experience: not recommended without assistance

Potential friction points:

• The feed trimming behavior (50 entry cap, 30-day window) means that if you accumulate a backlog — which is exactly what read-later tools encourage — items will silently vanish. The README explains the rationale (limiting feed size for RSS reader performance) but it’s a hard limit with no configuration option in the current version [README].
• Android integration requires manually configuring the HTTP Request Shortcuts app using the wiki, not a packaged app. It works, but it’s not polished [README].
• The browser extension lives in a separate repository (wezm/feedlynx-ext). At 107 stars on the main project, the extension is an even smaller community surface — verify it’s maintained before depending on it [README].
• There is no authentication for the feed URL itself beyond the token embedded in the path. The feed token in the URL is effectively your only protection against someone discovering and subscribing to your reading queue. It’s fine for personal use; it’s not enterprise-grade [README].

---

Pros and Cons

Pros

• Single binary, zero database. Feedlynx is as simple as a self-hosted tool gets: one Rust binary, one XML file on disk, two env vars. Nothing to migrate, nothing to back up except the XML file [README].
• RSS-native workflow. If you live in an RSS reader, your read-later queue lives in the same place. One less app to check [README][5].
• Cross-platform binaries. Pre-built for FreeBSD, Linux (x86/ARM), macOS Universal, and Windows. You don’t need a Rust toolchain to run it [README].
• YouTube embed support. Saves a YouTube link and gets an embeddable player entry rather than a bare URL — a small touch that matters if you save a lot of video content [README].
• Dual Apache-2.0 / MIT license. As permissive as open source gets [README].
• No vendor risk. The entire product is a file on disk and a small binary. There is no service to be shut down, no pricing change to absorb, no account to manage [README].
• Apple Shortcuts integration. iOS share sheet → Feedlynx works out of the box via a published Shortcut, which is the right integration for Apple users [README].

Cons

• Hard 50-entry cap. The feed trims aggressively. If you save faster than you read (common), links will disappear before you get to them with no warning [README]. This is not configurable in the current version.
• No permanent archival. Once a link ages out of the feed, it’s gone from Feedlynx’s perspective. No archive, no search history [README].
• No reading interface. This sounds obvious given the architecture, but it means annotations, highlights, and estimated read time — features people use daily in Pocket and Readwise — are simply absent [README].
• 107 stars, one maintainer. This is a one-person side project. There are 7 forks. If Wesley Moore stops maintaining it, the project stalls. The bus factor is 1 [README][1].
• Android is a manual integration. Not a first-class app — you configure a third-party HTTP shortcuts tool by hand [README].
• Feed URL security is token-only. The feed itself has no authentication beyond an obscure URL. Anyone with the URL can subscribe to your queue [README].
• No third-party reviews. The only coverage is directory listings [1][2][5]. There are no user testimonials, no head-to-head comparisons, no community threads to draw from. You’re largely flying on your own assessment.

---

Who should use this / who shouldn’t

Use Feedlynx if:

• You already run an RSS reader (Miniflux, FreshRSS, NetNewsWire, Reeder) and you want read-later to live in the same inbox.
• You’re comfortable running a binary behind a reverse proxy, or already have that infrastructure.
• You want permanent self-sovereignty over your saved links with no risk of a service shutdown.
• Your read-later habit is “save a few things per week and read them within a few days” — not “save 300 things and work through the backlog over months.”
• You’re replacing Pocket or Omnivore after those services closed or shut down.

Skip it (use Wallabag instead) if:

• You want full-text archival, tagging, search across your history, and a proper reading view.
• You prefer a web UI over an RSS reader workflow.
• You accumulate large backlogs and can’t trust a 50-entry trimming cap.

Skip it (use Readwise Reader) if:

• You want highlights, annotations, AI summaries, and a polished mobile reading experience and you’re willing to pay $7.99/mo for it.

Skip it (use Shiori or Linkwarden) if:

• What you actually want is a bookmark manager with tagging and search, not a read-later queue.

Skip it entirely if:

• You’re non-technical and don’t have someone to help with the reverse-proxy setup.
• You save links infrequently enough that the Instapaper free tier handles it.

---

Alternatives worth considering

• Wallabag — The self-hosted Pocket clone. PHP-based, full-text archival, mobile apps, permanent history, tagging. Considerably more infrastructure than Feedlynx but a complete replacement for what Pocket was. The obvious alternative if you want features [selfh.st directory].
• Miniflux + starring — If you’re already running Miniflux, its starring/bookmark feature is a lightweight read-later queue without a separate tool.
• Shiori — Self-hosted bookmark manager. Closer to del.icio.us than to Pocket — designed for permanent archival and organization, not a temporary queue.
• Linkwarden — Self-hosted link manager with collections, tagging, and full-page archival. Active development, 10k+ stars. More overhead than Feedlynx but significantly more capable [selfh.st newsletter referenced a Linkwarden widget update in the same issue that listed Feedlynx [1]].
• Readwise Reader — The best-in-class paid option if you read long-form regularly and want annotations, AI summaries, and newsletter integration. $7.99/mo is the honest comparison.
• Omnivore — Was open-source, shut down 2024. Mentioned here only because many Feedlynx users are Omnivore refugees.
• ntfy + RSS reader bookmarking — DIY equivalent for the technically inclined: send URLs to yourself via ntfy or similar, consume in your reader. More work, no polished client integrations.

---

Bottom line

Feedlynx is a deliberately narrow tool that gets one thing exactly right: if you live in an RSS reader, your read-later queue should live there too, not in a separate app. The architecture is sensible, the binary is small, the setup is straightforward for anyone running a homelab, and the Apache-2.0/MIT dual license means no strings attached.

The limitations are real and honest: 107 stars and one maintainer means you’re betting on a small side project; the 50-entry feed cap will surprise heavy savers; and there’s no web interface, no archival, and no annotations. This isn’t trying to be Wallabag or Readwise Reader. It’s trying to be a small, reliable piece of Unix-style infrastructure that does one job and stays out of your way. For RSS users who want exactly that, it delivers. For everyone else, the feature gap is real and the alternatives are better suited.

If you’re replacing Pocket after its shutdown, or just tired of paying for a service that might fold, the self-hosting math here is as simple as it gets: one binary, one file, one less subscription.

---

Sources

1. selfh.st — This Week in Self-Hosted (16 August 2024). Feedlynx listed in “Latest Directory Apps” as new software. https://selfh.st/weekly/2024-08-16/
2. Adam Douglas — Little Bits: Issue #15 (adamsdesk.com, published Aug 1, 2024). Link roundup mentioning Feedlynx. https://www.adamsdesk.com/posts/little-bits-issue-15/
3. garrido.io — Microblog. RSS/self-hosting practitioner blog, context on the self-hosted RSS reader community. https://garrido.io/microblog/
4. selfhostsource.com — Software Directory. Self-hosted software directory listing. https://selfhostsource.com/software/
5. blog.jutty.dev — Links (Oct 26, 2025 entry). Feedlynx listed as “Collect links to read later into an RSS feed by Wesley Moore (wezm)”. https://blog.jutty.dev/links/

Primary sources:

• GitHub repository and README: https://github.com/wezm/feedlynx (108 stars, dual Apache-2.0/MIT license, maintained by Wesley Moore)
• crates.io package: https://crates.io/crates/feedlynx
• Browser extension repository: https://github.com/wezm/feedlynx-ext