steipete/summarize

What it solves

Summarize simplifies the process of extracting clean, readable text from various noisy sources—such as web pages, PDFs, images, and audio/video files—and converting that content into concise summaries using LLMs.

How it works

The tool uses a multi-path extraction system: it employs readability extraction for web pages (with Markdown and Firecrawl fallbacks), handles file-aware extraction for documents, and retrieves transcripts for YouTube and podcasts. For media without transcripts, it uses local or cloud transcription and can perform OCR on video slides. These extracted texts are then passed to a configured LLM provider (such as OpenAI, Ollama, or OpenRouter) to generate the final summary.

Who it’s for

It is designed for users who want a quick way to distill information from the web or media files, whether via a Node.js CLI, a browser extension (Chrome/Firefox), or programmatically through a core library.

Highlights

• Versatile Input Support: Handles web pages, PDFs, images, YouTube videos, podcasts, and raw audio/video files.
• Flexible Model Integration: Works with various API providers, OpenAI-compatible endpoints, local models via Ollama, and authenticated coding CLIs like GitHub Copilot and Claude.
• Visual Extraction: Capable of extracting scene-based frames from videos with optional OCR for slides.
• Multiple Interfaces: Available as a global CLI, a browser side-panel extension, and a Node.js library.