Most summarizers lock you into one model behind one hosted API. This tool's organizing idea is the opposite: the backend is a swappable slot. Point it at a local coding CLI you already run — Claude Code, Codex, Gemini — or at a hosted API provider like OpenAI, Google, or xAI, and the same content pipeline feeds whichever you pick.
What Sets It Apart
- Backend-agnostic by design: eight local CLI backends and a comparable roster of hosted API providers — OpenAI, Anthropic, Google, xAI, OpenRouter, OpenAI-compatible local endpoints — are all first-class, so switching models is a config change rather than a rewrite.
- Handles far more than text: YouTube and podcast RSS feeds plus audio and video transcription via Whisper. Web pages go through a direct fetch-and-extract pipeline first, falling back to Firecrawl (
--firecrawl auto) only when a page is blocked or too thin; PDFs and other files are inlined into the model prompt, so quality depends on the provider (Google is most reliable here). - Video slides extraction is the standout: it captures screenshots, runs OCR, and emits timestamped cards, turning a long talk into a skimmable deck instead of a wall of transcript.
- Two front ends share one core: a Chrome side panel / Firefox sidebar with streaming chat and history, plus a CLI for scripted or batch runs.
Who It's For
Great fit if you want summarization that plugs into model access you already have — a coding-CLI subscription or an API key — or if you routinely digest video and audio, not just articles. The browser extension's Direct mode runs on Gemini Nano with no key for quick web summaries, but the full pipeline is a different story: rich media needs FFmpeg and Whisper installed, and CLI backends need their binaries authenticated, so look elsewhere if you want everything to work with zero local setup.
