Caveman
Compacts text, URLs, files, and various data into highly condensed classical Chinese (Wenyan) using a two-pass mechanical and LLM-driven compression pipeline.
Compacts text, URLs, files, and various data into highly condensed classical Chinese (Wenyan) using a two-pass mechanical and LLM-driven compression pipeline.
Caveman is an MCP server designed to dramatically reduce text size through a unique two-pass compression method. It first performs a mechanical pass to drop filler words and preserve critical elements like code blocks and numbers, then utilizes an LLM to convert the text into highly concise classical Chinese (Wenyan). The tool offers specialized functions for extracting and compressing content from diverse sources, including smart routing for various URL patterns (YouTube, GitHub, Arxiv, Reddit, RSS), comprehensive file type support (images, audio, documents, spreadsheets), and dedicated processing for Git artifacts (diffs, logs) and error logs, achieving typical compression ratios of 65-80% while preserving essential information.