Lean-ctx

•
4.0k
•

Optimizes AI coding workflows by transparently reducing LLM token consumption up to 99% through context compression and an intelligent server.

Lean-ctx is designed to drastically cut LLM token costs for developers using AI coding assistants such as Cursor, Claude Code, GitHub Copilot, Windsurf, and Gemini CLI. It operates transparently between your AI tool and the large language model, implementing a multi-pronged approach that includes a shell hook for compressing CLI output (over 90 patterns), a sophisticated context server offering 28 intelligent tools for optimized context delivery, and AI tool-specific integration hooks. This innovative system compresses code and command outputs by up to 99%, making AI coding significantly faster, more cost-effective, and efficient without requiring workflow changes.

主な機能

01491 GitHub stars
02Transparent Shell Hook compressing CLI output for 90+ commands and 34 categories.
03Context Server with 28 intelligent tools for cached reads, multi-agent sharing, and adaptive context modes.
04Achieves up to 99% reduction in LLM token consumption for AI coding sessions.
05One-command integration with major AI coding tools including Cursor, Claude Code, and Copilot.
06Zero telemetry for enhanced privacy and a single Rust binary for lightweight operation.

ユースケース

01Reducing token costs for AI coding assistants like Claude Code, Cursor, and Copilot.
02Optimizing context delivery to LLMs for faster and cheaper AI-driven software development.
03Setting up an efficient, open-source MCP server to manage LLM interactions for developer tools.