AiCode - CLI Coding Assistant

A resource-efficient CLI coding assistant optimized for small local models with full tool execution capabilities.

Features

🤖 Multi-Model Support: Works with Ollama, LM Studio, and any OpenAI-compatible endpoint 🛠️ Tool Execution: File operations, terminal commands, code analysis, and image processing 🖼️ Image Analysis: Visual debugging and understanding capabilities 💬 Interactive Shell: Rich CLI interface with command completion and history 🔧 Optimized for Small Models: Efficient prompts and context management 🛡️ Safety Features: Safe mode with confirmation for dangerous operations

Quick Start

Installation

pip install click rich requests pillow prompt-toolkit

Run AiCode

python main.py

Basic Commands

# Start interactive shell
python main.py

# Configure settings
python main.py configure

# List available models
python main.py models

# Ask a single question
python main.py ask "write a function to sort a list"

Configuration

Default model endpoints are configured in config.ini:

• Ollama: http://localhost:11434/api/chat
• LM Studio: http://localhost:1234/v1/chat/completions
• Default: http://localhost:8080/v1/chat/completions

Interactive Shell Commands

Command	Description
/help	Show available commands
/models	List configured models with status
/switch <model>	Switch to different model
/tools	Show available tools
/clear	Clear conversation history
/image <path>	Analyze image file
/analyze <file>	Analyze code file
/debug	Enter debug mode
/exit	Exit AiCode

Tool Capabilities

File Operations

• Read/write files with syntax highlighting
• Directory listing and creation
• File deletion (with safety confirmation)
• Path validation and existence checking

Terminal Operations

• Safe command execution with timeouts
• Working directory management
• Output capture and formatting

Code Analysis

• Syntax error detection
• Code structure analysis
• Basic formatting capabilities
• Language-specific parsing (Python AST)

Image Analysis

• Visual debugging support
• Image optimization for small models
• Basic visual analysis (colors, dimensions)
• Image comparison functionality

Usage Examples

Basic Conversation

aicode> write a function to calculate fibonacci numbers

File Operations

aicode> read the file main.py and analyze it
aicode> TOOL: read_file(path="main.py")

Image Analysis

aicode> /image screenshot.png
aicode> analyze this error screenshot

Model Switching

aicode> /switch ollama
aicode> /models

Safety Features

• Safe Mode: Enabled by default, asks for confirmation on dangerous operations
• Path Validation: Prevents unauthorized file access
• Command Filtering: Blocks potentially destructive terminal commands
• Timeout Protection: Prevents hanging operations

Small Model Optimization

• Context Compression: Automatically truncates context to fit token limits
• Specialized Prompts: Optimized for coding, debugging, and analysis tasks
• Efficient Processing: Minimal memory footprint and fast response times
• Image Optimization: Resizes images for optimal model processing

Configuration Options

[DEFAULT]
max_tokens = 2048
temperature = 0.7
context_length = 4096
timeout = 30

[models]
default = http://localhost:8080/v1/chat/completions
ollama = http://localhost:11434/api/chat
lmstudio = http://localhost:1234/v1/chat/completions

[tools]
enable_file_operations = true
enable_terminal_commands = true
enable_image_analysis = true
safe_mode = true

Context Management

AiCode automatically loads system context from aicode.md which provides comprehensive information about capabilities and usage patterns. This enables the AI to understand its environment and provide better assistance.

Development

The project structure is modular and extensible:

aicode/
├── __init__.py          # Package initialization
├── cli.py              # CLI interface and commands
├── config.py           # Configuration management
├── models.py           # Model communication
├── tools.py            # Tool execution system
├── prompts.py          # Prompt optimization
├── shell.py            # Interactive shell
└── image_analyzer.py   # Image processing

Requirements

• Python 3.7+
• Local model server (Ollama, LM Studio, etc.)
• Terminal with basic cursor support

License

Open source - feel free to modify and extend for your needs.