01Compresses logs into LLM-readable structured text, preserving critical information and highlighting anomalies.
02Achieves 40-60% token savings for LLMs by identifying and abstracting repetitive log patterns.
03Provides both a high-performance Rust CLI and a user-friendly Python API for versatile integration.
04Integrates as an MCP Server for Claude Code, enabling direct LLM interaction for server-side log analysis.
05Utilizes an advanced compression pipeline with normalization, frequency analysis, and multi-pass BPE for optimal results.
063 GitHub stars