Feed reader with an LLM sidekick and a Rust backbone

What it does Saga Reader is a desktop app that lets you subscribe to topics rather than just feeds, then scrapes the web via search and RSS to pull in relevant articles. Everything is stored locally in SQLite, summarized by cloud or local LLMs, and consumed through a built-in reader—or an AI companion that will chat about the article in real time. It is essentially a private, self-hosted newsroom with an opinionated librarian.

The interesting bit The backend is a Rust monorepo with discrete crates for scraping, LLM abstraction, local Ollama hosting, and storage, all wired to a Svelte frontend through Tauri. That modularity means you could theoretically swap search providers or models without touching the UI, though right now Bing and GLM Flash appear to be the primary wired-up options.

Key highlights

• Claims to run under 10 MB of RAM, targeting older hardware.
• All data stays on your machine; no ads or third-party tracking.
• Supports both cloud LLMs and local models via Ollama.
• Automatically translates foreign-language articles into your preferred language.
• Background updates and a built-in article viewer keep the workflow inside one window.

Caveats

• The contribution guide contains a stray reference to an unmentioned “wunderlist,” suggesting the docs are still being cleaned up.
• Out-of-the-box integrations are narrow: Bing is the default search engine and GLM Flash is the primary cloud model mentioned, with Google and OpenAI still listed as to-do items.

Verdict Try it if you want a lightweight, private alternative to cloud news aggregators and don’t mind managing your own API endpoints or a local Ollama instance. Pass if you need a mature reader with broad search and model support on day one.