lean-ctx

Overview

lean-ctx is a local-first context runtime that functions as both an MCP server and shell hook. It compresses file reads and shell output before they reach the LLM, with cached re-reads dropping to approximately 13 tokens. The project provides 51 tools, 10 read modes, and 60+ patterns to optimize context management across multiple AI coding tools like Cursor, Claude Code, and Copilot. It features a property graph for code intelligence, session memory capabilities, and context packaging for sharing across projects.

Try asking AI

After installing, here are 5 things you can ask your AI assistant:

• ctx_refactor

  Language-server-powered refactoring operations (rename, references, go-to-definition, find-implementations)

  Try:Rename this function throughout the codebaseTry:Find all references to this variableTry:Show me where this interface is implemented
• ctx_expand

  Archive full-text search over all previously archived tool outputs

  Try:Search for all previous mentions of user authenticationTry:Find past conversations about API performanceTry:Search the archive for database connection issues
• ctx_read

  Cached + mode-aware file reads with multiple options

  Try:Read this file in map mode to see its structureTry:Get the signatures of all functions in this fileTry:Show me the diff of this file compared to the last commit
• ctx_pack

  Create context packages bundling knowledge, graph, session, and gotchas

  Try:Create a context package for this feature branchTry:Pack the current session into a shareable .lctxpkg fileTry:Create a package with all related files for this bug
• ctx_search

  Search across codebase with hybrid search capabilities

  Try:Find all functions that handle user authenticationTry:Search for similar code patterns across the projectTry:Find all implementations of this trait
• ctx_memory

  Persist and query session facts across multiple conversations

  Try:What decisions did we make about database schema?Try:Recall previous work we did on authentication flowTry:What were the key findings from our last discussion?
• ctx_graph

  Query the property graph for code relationships and impact analysis

  Try:Show me the impact of changing this functionTry:Find files that are related to this moduleTry:What imports are affected by this change?
• ctx_benchmark

  Benchmark and measure compression effectiveness by language and mode

  Try:Run a benchmark on the current repositoryTry:Show compression metrics for TypeScript filesTry:Compare token savings across different read modes
• ctx_inspect

  Inspect repository structure and configuration

  Try:Show me the project configurationTry:List all supported file types in this repositoryTry:What compression patterns are active for this project?
• ctx_export

  Export context data for external tools or sharing

  Try:Export current context to a JSON fileTry:Prepare context for external analysis toolTry:Share session context with a teammate
• ctx_import

  Import external context or previous sessions

  Try:Import context from a previous session fileTry:Load context package from a .lctxpkg fileTry:Restore context from backup
• ctx_analyze

  Analyze code patterns and provide insights

  Try:Analyze code quality issues in this fileTry:Identify potential performance bottlenecksTry:Check for security vulnerabilities in this code

Comparable tools

Installation

# Universal installer (no Rust needed)
curl -fsSL https://leanctx.com/install.sh | sh

# macOS / Linux via Homebrew
brew tap yvgude/lean-ctx && brew install lean-ctx

# Node.js
npm install -g lean-ctx-bin

# Rust
cargo install lean-ctx

# Pi Coding Agent
pi install npm:pi-lean-ctx

lean-ctx setup

{
  "mcpServers": {
    "lean-ctx": {
      "command": "lean-ctx",
      "args": ["mcp"]
    }
  }
}

FAQ

Which AI coding tools does lean-ctx support?

lean-ctx supports over 20 AI coding tools including Cursor, Claude Code, GitHub Copilot, Windsurf, Gemini, and many others. It works with any MCP-compatible client and has specific integrations for various IDEs like VS Code, Neovim, and JetBrains.

How does lean-ctx reduce token usage?

lean-ctx reduces token usage by caching file reads with compression (down to ~13 tokens for cached re-reads), compressing shell output using 60+ patterns, and using intelligent read modes that only send relevant context based on the task.

Compare lean-ctx with