summarize-anything

What problem does it solve?

Manually summarizing long-form content like transcripts, articles, and documents is tedious, and standard LLMs fail when input exceeds their context window. This Skill eliminates that bottleneck by handling text of any length (1k to 1M+ words) and repurposing it into dozens of usable formats without manual copy-pasting between tools.

Core Features & Use Cases

• Recursive map-reduce summarization: Automatically chunks long text on paragraph boundaries with overlap, summarizes each chunk, then synthesizes results for inputs up to 1M+ words.
• Pluggable LLM backends: Works with OpenAI, Anthropic, Gemini, Ollama, OpenRouter, or any OpenAI-compatible endpoint, with automatic backend selection based on input size.
• 17 output formats: Generates executive summaries, YouTube descriptions, tweet threads, blog outlines, thumbnail prompts, and more from a single input.
• Focus directives: Steer summary emphasis to highlight specific angles (e.g., "focus on the technical architecture decisions" or "emphasize the personal story").
• Use case: A content creator with a 2-hour podcast transcript can use this Skill to generate a YouTube description, chapter markers, 3 tweet options, a LinkedIn post, and a blog outline in one workflow, instead of running separate prompts for each format.

Quick Start

Use the summarize-anything skill to turn the attached 2-hour podcast transcript into a YouTube description, chapter markers, and a tweet thread, emphasizing the guest's AI cancer treatment research.