LeanCTX is a hybrid context optimizer designed to reduce token consumption
in AI-assisted software engineering workflows. It acts as a cognitive filter
between AI tools and Large Language Models (LLMs), stripping noise like
comments and whitespace to lower costs and maximize context efficiency.

Features:
- AST-Aware Compression: Uses Tree-sitter to parse 14 languages, preserving
  critical types and logic while reducing token counts by up to 90%.
- MCP Server Integration: Provides 24 tools that replace standard file reads
  with compressed alternatives for any MCP-enabled AI client.
- Automated Shell Hook: Intercepts and compresses output from over 90 common
  command patterns (git, npm, cargo, docker, kubectl, etc.).
- Privacy and Performance: Single Rust binary with zero dependencies and no
  cloud requirements.