LeoCoder 🚀

Smart LLM router for vibe coding - local first, cloud fallback

Save cloud tokens by intelligently routing simple tasks to local models and complex tasks to cloud APIs.

Features

• 🏠 Local-First: Automatically uses local models (Ollama, LM Studio) for simple tasks
• ☁️ Cloud Fallback: Routes complex tasks to cloud providers (Groq, Gemini)
• 🧠 Smart Routing: AI-powered task classification based on complexity
• 💰 Token Savings: Track tokens saved by using local models
• 🛠️ File Operations: Read, write, edit files directly from chat
• 💻 Shell Commands: Run terminal commands safely
• 🔍 Code Search: Grep-like search across your codebase

Installation

1. Clone and Install Dependencies

cd LeoCoder
npm install
npm run build

2. Setup Environment Variables

Copy the example env file and add your API keys:

copy .env.example .env

Edit .env and add your keys:

# Get from https://console.groq.com/keys (Free tier)
GROQ_API_KEY=gsk_...

# Get from https://makersuite.google.com/app/apikey (Free tier)
GEMINI_API_KEY=...

3. Install Local Models (Recommended)

For Ollama (recommended for low-end hardware):

# Install Ollama from https://ollama.ai

# Download a lightweight coding model
ollama pull phi-3-mini
# or
ollama pull qwen2.5-coder:1.5b
# or
ollama pull deepseek-coder:1.3b

For LM Studio:

1. Download from https://lmstudio.ai
2. Download a quantized model (1-3B recommended)
3. Start the local server

Usage

Start Interactive Chat

npm start
# or
node dist/index.js chat

Ask a Single Question

node dist/index.js ask "Explain this function..."

Check Provider Status

node dist/index.js status

Chat Commands

Command	Description
/file <path>	Open a file for context
/read <path>	Read and display a file
/search <pattern>	Search for pattern in files
/run <command>	Run a shell command
/models	Show active models
/stats	Show token usage stats
/help	Show help
exit, quit	End session

How Routing Works

The router analyzes each request and assigns a complexity score (1-10):

Complexity	Route	Example Tasks
1-3	🏠 Local	Autocomplete, explain code, format, simple fixes
4-6	🏠 Local (default)	Debug, optimize, small refactors
7-10	☁️ Cloud	Architecture, complex refactors, security audits

Examples

You: "complete this function"
📍 Routing: LOCAL (Complexity: 2/10, Task: code_completion)

You: "design a microservice architecture for..."
📍 Routing: CLOUD (Complexity: 8/10, Task: architecture_design)

You: "explain what this code does"
📍 Routing: LOCAL (Complexity: 3/10, Task: explain_code)

Configuration

Edit config.yaml to customize:

local:
  preferred_models:
    - phi-3-mini      # Best for coding
    - qwen2.5-coder:1.5b
    - deepseek-coder:1.3b
  max_model_size_gb: 4  # Adjust for your RAM

cloud:
  token_budget:
    daily_limit: 100000  # Cloud tokens per day

routing:
  complexity:
    local_threshold: 3   # Below = local
    cloud_threshold: 7   # Above = cloud

Recommended Models for Low-End Hardware

For i5-8th Gen with limited RAM:

Model	Size	Quality	Speed
phi-3-mini	3.8B	⭐⭐⭐⭐	Fast
qwen2.5-coder:1.5b	1.5B	⭐⭐⭐	Very Fast
deepseek-coder:1.3b	1.3B	⭐⭐⭐	Very Fast
tinyllama	1.1B	⭐⭐	Fastest

Pull Models

ollama pull phi-3-mini
ollama pull qwen2.5-coder:1.5b
ollama pull deepseek-coder:1.3b

Token Savings Estimate

Based on typical usage:

Task Type	Local %	Cloud %	Est. Savings
Autocomplete	100%	0%	~50k tokens/day
Code Explanation	90%	10%	~30k tokens/day
Debugging	60%	40%	~15k tokens/day
Complex Tasks	0%	100%	-

Total estimated savings: 70-80% of cloud tokens

Troubleshooting

"No local provider available"

• Make sure Ollama is running: ollama serve
• Check LM Studio server is started
• Verify base URLs in config

"Cloud provider unavailable"

• Check API keys in .env
• Verify internet connection
• Check API quota limits

Model too slow

• Use smaller models (1-3B)
• Enable CPU-only mode
• Close other applications

License

MIT