Houtini LM is an MCP server that connects Claude to LM Studio for local AI-powered development without API costs.
Core Capabilities:
• Code Analysis - Quality assessment, security audits, dependency analysis, project structure evaluation, execution path tracing, and OWASP compliance checking • Code Generation - Unit test creation, JavaScript to TypeScript conversion, responsive component generation, and refactoring suggestions • WordPress Development - Complete plugin/theme generation, security audits, and static site to WordPress conversion • Creative Projects - CSS art generation, HTML5 arcade games, and interactive text adventures • Documentation - Professional documentation generation from code analysis in multiple formats • Performance Optimization - Unused file/CSS detection, database query analysis, and workflow optimization • System Utilities - Health checks, function discovery, cache management, and path resolution • Custom Prompting - Universal fallback for any analysis or generation task with dynamic token allocation
The server acts as an unlimited local AI companion, offloading demanding tasks from Claude while providing comprehensive development tools for both technical analysis and creative projects.
Generates pure CSS art and animations through creative project functions
Creates complete playable HTML5 games and responsive HTML components with accessibility features
Performs deep code analysis, quality assessment, security auditing, and generates comprehensive unit tests for JavaScript files
Generates professional documentation and beautiful project structure trees in markdown format
Analyzes Node.js projects including dependency analysis, circular dependency detection, and unused import identification
Conducts security audits and vulnerability scanning with OWASP compliance checking
Generates comprehensive unit tests and analyzes React components for quality and best practices
Converts JavaScript files to TypeScript with strict mode and type safety implementation
Creates complete WordPress plugins with custom post types and admin interfaces, and performs security audits on WordPress themes
Houtini LM - LM Studio MCP Server with Expert Prompt Library and Custom Prompting
Your unlimited AI companion: This MCP server connects Claude to LM Studio for code analysis, generation, and creativity
Transform your development workflow with our expert-level prompt library for code analysis, professional documentation generation, and creative project scaffolding - all running locally without API costs. For developers, vibe coders and creators alike.
What This Does
Houtini LM saves your Claude context window by offloading detailed analysis tasks to LM Studio locally or on your company network whilst Claude focuses on strategy and complex problem-solving. Think of it as your intelligent coding assistant that never runs out of tokens.
Perfect for:
🔍 Code analysis - Deep insights into quality, security, and architecture
📝 Documentation generation - Professional docs from code analysis
🏗️ Project scaffolding - Complete applications, themes, and components
🎮 Creative projects - Games, CSS art, and interactive experiences
🛡️ Security audits - OWASP compliance and vulnerability detection
Quick Start Prompt Guide
Once installed, simply use natural language prompts with Claude:
Prerequisites
Essential Requirements:
LM Studio - Download from lmstudio.ai
Must be running at
ws://127.0.0.1:1234
Model loaded and ready (13B+ parameters recommended)
Desktop Commander MCP - Essential for file operations
Repository: DesktopCommanderMCP
Required for reading files and writing generated code
Node.js 24.6.0 or later - For MCP server functionality
Download from nodejs.org
Claude Desktop - For the best experience
Download from claude.ai/download
Installation
1. Install the Package
2. Configure Claude Desktop
Add to your Claude Desktop configuration file:
Windows: %APPDATA%/Claude/claude_desktop_config.json
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
3. Start LM Studio
Launch LM Studio
Load a model (13B+ parameters recommended for best results)
Start the server at
ws://127.0.0.1:1234
Verify the model is ready and responding
4. Verify Installation
Restart Claude Desktop, then test with:
Available Functions
🔍 Analysis Functions (17 functions)
analyze_single_file
- Deep code analysis and quality assessmentcount_files
- Project structure with beautiful markdown treesfind_unused_files
- Dead code detection with risk assessmentsecurity_audit
- OWASP compliance and vulnerability scanninganalyze_dependencies
- Circular dependencies and unused importsAnd 12 more specialized analysis tools...
🛠️ Generation Functions (10 functions)
generate_unit_tests
- Comprehensive test suites with framework patternsgenerate_documentation
- Professional docs from code analysisconvert_to_typescript
- JavaScript to TypeScript with type safetygenerate_wordpress_plugin
- Complete WordPress plugin creationgenerate_responsive_component
- Accessible HTML/CSS componentsAnd 5 more generation tools...
🎮 Creative Functions (3 functions)
css_art_generator
- Pure CSS art and animationsarcade_game
- Complete playable HTML5 gamescreate_text_adventure
- Interactive fiction with branching stories
⚙️ System Functions (5 functions)
health_check
- Verify LM Studio connectionlist_functions
- Discover all available functionsresolve_path
- Path analysis and suggestionsAnd 2 more system utilities...
Context Window Management
Houtini LM implements intelligent context window management to maximize the efficiency of your local LM models while ensuring reliable processing of large files and complex analysis tasks.
Dynamic Context Allocation
Adaptive Context Utilization: Unlike systems with hardcoded token limits, Houtini LM dynamically detects your model's context window and allocates 95% of available tokens for optimal performance:
Benefits:
✅ Maximum efficiency - No wasted context space
✅ Model-agnostic - Works with any context size (4K, 16K, 32K+)
✅ Future-proof - Automatically adapts to larger models
Three-Stage Prompt System
Houtini LM uses a sophisticated prompt architecture that separates concerns for optimal token management:
Stage 1: System Context - Expert persona and analysis methodology
Stage 2: Data Payload - Your code, files, or project content
Stage 3: Output Instructions - Structured response requirements
Intelligent Processing:
Small files → Single-stage execution for speed
Large files → Automatic chunking with coherent aggregation
Multi-file projects → Optimized batch processing
Automatic Chunking Capability
When files exceed available context space, Houtini LM automatically chunks content while maintaining analysis quality:
Smart Chunking Features:
🔍 Natural boundaries - Splits at logical sections, not arbitrary points
🔄 Context preservation - Maintains analysis continuity across chunks
📊 Intelligent aggregation - Combines chunk results into coherent reports
⚡ Performance optimization - Parallel processing where possible
Example Chunking Process:
Timeout Configuration
Houtini LM uses 120-second timeouts (2 minutes) to accommodate thorough analysis on lower-powered systems:
Why Extended Timeouts:
🔍 Complex analysis - Security audits, architecture analysis, and comprehensive code reviews take time
💻 System compatibility - Works reliably on older hardware and resource-constrained environments
🧠 Model processing - Larger local models (13B-33B parameters) require more inference time
📊 Quality over speed - Comprehensive reports are worth the wait
Timeout Guidelines:
Simple analysis (100 lines): 15-30 seconds
Medium files (500 lines): 30-60 seconds
Large files (1000+ lines): 60-120 seconds
Multi-file projects: 90-180 seconds
Performance Tips:
Use faster models (13B vs 33B) for quicker responses
Enable GPU acceleration in LM Studio for better performance
Consider using
analysisDepth="basic"
for faster results when appropriate
Memory Efficiency
Intelligent Caching: Results are cached to prevent redundant processing
Resource Management: Automatic cleanup of large contexts after processing
Streaming Responses: Progressive output delivery for better user experience
This architecture ensures Houtini LM can handle everything from small utility functions to entire enterprise codebases while maintaining consistent quality and performance across different hardware configurations.
Documentation
Complete guides available:
Analysis Functions Guide - All 17 analysis tools
Generation Functions Guide - All 10 creation tools
Creative Functions Guide - Games and art tools
System Functions Guide - Utilities and diagnostics
Complete User Guide - Comprehensive usage manual
Recommended Setup
For Professional Development:
CPU: 8-core or better (for local LLM processing)
RAM: 32GB (24GB for model, 8GB for development)
Storage: SSD with 100GB+ free space
Model: Qwen2.5-Coder-14B-Instruct or similar
Performance Tips:
Use 13B+ parameter models for professional-quality results
Configure
LLM_MCP_ALLOWED_DIRS
to include your project directoriesInstall Desktop Commander MCP for complete file operation support
Keep LM Studio running and model loaded for instant responses
Version History
Version 1.0.0 (Current)
✅ Complete function library (35+ functions)
✅ Professional documentation system
✅ WordPress-specific tools and auditing
✅ Creative project generators
✅ Comprehensive security analysis
✅ TypeScript conversion and test generation
✅ Cross-file integration analysis
License
MIT License - Use this project freely for personal and commercial projects. See LICENSE for details.
Contributing
We welcome contributions! Please see our Contributing Guidelines for details on:
Code standards and patterns
Testing requirements
Documentation updates
Issue reporting
Support
Issues: GitHub Issues
Discussions: GitHub Discussions
Documentation: Complete guides in the
docs/
directory
Ready to supercharge your development workflow? Install Houtini LM and start building amazing things with unlimited local AI assistance.
Built for developers who think clearly but can't afford to think expensively.
local-only server
The server can only run on the client's local machine because it depends on local resources.
Tools
Houtini LM - LM Studio MCP Server with Expert Prompt Library and Custom Prompting - Offload tasks to LM Studio from Claude Desktop.