Self-hosted RSS curation: translate, filter, merge

What it does

RSSBox is a self-hosted Django application that ingests RSS feeds, runs them through translation and filtering pipelines, and emits new RSS or JSON feeds you can subscribe to in any reader. It fetches full text, supports bilingual output, and lets you merge multiple sources into a single tagged feed. Think of it as a preprocessor for your reader that handles the chores most readers skip.

The interesting bit

The project treats translation as infrastructure: each source gets its own engine—DeepL or any OpenAI-compatible model—and it tracks token or character spend per feed so you know exactly what multilingual curiosity costs you. The AI filtering and keyword filtering happen upstream, which means you can subscribe to high-volume sources without drowning.

Key highlights

• Per-source translation engine selection (DeepL, OpenAI-compatible APIs)
• Bilingual display mode and full-text fetching
• Keyword and AI-based filtering to cut noise before delivery
• Tag-based feed merging: bundle multiple sources into one unified RSS/JSON feed
• Token and character cost tracking per source
• Self-hosted with Docker support; formerly known as RSS Translator

Caveats

• The README lists features but stays quiet on how the AI filtering actually works or which models it uses.
• Architecture details beyond Django and Docker badges are left to the source code.

Verdict

Worth a look if you read across languages or manage noisy, high-volume feeds and prefer reader-agnostic preprocessing. Skip it if your current reader already handles translation and filtering well enough.