LeanKG

Overview

LeanKG is a Rust-based local knowledge graph designed to provide AI coding tools with accurate codebase context. It indexes code files, builds dependency graphs, and exposes an MCP server that allows AI tools like Cursor, OpenCode, and Claude Code to query the knowledge graph directly. The system offers features such as impact radius calculation, dependency graph visualization, token optimization, and multi-language support. Unlike cloud-based solutions, LeanKG operates locally without external database dependencies, addressing the high token consumption problem when AI tools process full codebase context.

Try asking AI

After installing, here are 7 things you can ask your AI assistant:

• search_code

  Search for code patterns or specific implementations in the knowledge graph

  Try:Find all functions that process user authenticationTry:Search for SQL injection patterns in the codebase
• find_dependencies

  Find dependencies between code elements and analyze impact radius

  Try:Show all dependencies for the main functionTry:What would be affected if I change this API endpoint?
• get_element_info

  Retrieve detailed information about a specific code element

  Try:Show information about the PaymentProcessor classTry:Get details on the authenticate function
• detect_clusters

  Identify functional code communities and clusters in the codebase

  Try:Find all related authentication modulesTry:Show clusters of utility functions
• trace_feature

  Trace feature implementation across the codebase

  Try:Trace the user registration flow through the codeTry:Show all components that handle payment processing
• annotate_code

  Add annotations to code elements for better context

  Try:Mark the main function as an entry pointTry:Add documentation to the payment processing module
• export_graph

  Export the knowledge graph in various formats

  Try:Export the dependency graph as MermaidTry:Get the graph in JSON format for analysis
• analyze_quality

  Analyze code quality and find oversized functions

  Try:Find all functions larger than 100 linesTry:Identify potential code smells in the authentication module
• get_metrics

  View token savings and context reduction metrics

  Try:Show token usage statistics for this weekTry:How much context has been saved since yesterday?
• detect_changes

  Detect changes in codebase and classify risk level

  Try:Check for critical changes in the payment moduleTry:What risks are associated with recent database changes?
• run_command

  Execute shell commands with RTK compression to save tokens

  Try:Run tests with token compressionTry:Execute build process with context optimization
• register_repo

  Register a new repository for multi-repo context management

  Try:Add the frontend repository to LeanKGTry:Register the microservices repo

Comparable tools

Installation

curl -fsSL https://raw.githubusercontent.com/FreePeak/LeanKG/main/scripts/install.sh | bash -s -- <target>

cargo install leankg && leankg --version

# Install with Claude Code hooks
leankg setup
# Restart Claude Code or run:
/reload-plugins

FAQ

What AI tools does LeanKG support?

LeanKG supports Cursor, Claude Code, OpenCode, Kilo Code, Gemini CLI, Google Antigravity, and Codex with varying levels of integration including auto-setup, session hooks, and plugins.

How does LeanKG reduce token consumption?

Instead of processing full codebase context, LeanKG builds a knowledge graph that allows AI tools to query targeted subgraphs, reducing context tokens from 15,000-45,000 to focused relevant code.

Is LeanKG cloud-based?

No, LeanKG is a local-first solution that operates entirely on your machine without cloud services or external database dependencies.

Compare LeanKG with