Terminal Coding Assistant (gpp)

A powerful terminal-based AI coding assistant that uses LM Studio to help you create, read, and edit code files efficiently. Now with support for multiple file operations and global terminal access.

✨ Features

• Create - Generate new code files from natural language prompts
• Read - Analyze and explain existing code files
• Edit - Modify existing files based on instructions
• Multiple Files - Create, read, and edit multiple files in one command
• Model Management - List, load, and unload models from LM Studio
• Multi-language - Supports C, C++, Java, Python, HTML, CSS, PHP, JavaScript, TypeScript
• Global Access - Use gpp from any terminal directory after installation

🔧 Prerequisites

1. LM Studio running locally on http://localhost:1234
2. Python 3.6+ with requests library

   pip3 install requests

📦 Installation

Quick Install (Recommended)

Linux/macOS:

chmod +x install.sh
./install.sh

Windows (PowerShell):

.\install.ps1

Windows (Command Prompt):

install.bat

Manual Installation

Linux/macOS:

# Copy files to ~/.local/bin
mkdir -p ~/.local/bin
cp gpp ~/.local/bin/gpp
cp model_manager.py ~/.local/bin/model_manager.py
chmod +x ~/.local/bin/gpp

# Add to PATH (add to ~/.bashrc, ~/.zshrc, or ~/.profile)
export PATH="$HOME/.local/bin:$PATH"

Windows:

1. Create %USERPROFILE%\.local\bin directory
2. Copy gpp and model_manager.py to that directory
3. Add the directory to your system PATH
4. Restart terminal

✅ Verify Installation

./verify_setup.sh

This will check:

• ✅ gpp is accessible globally
• ✅ Python 3 is installed
• ✅ requests library is available
• ✅ LM Studio is running and accessible
• ✅ Models are available

🚀 Usage

Single File Operations

# Create a new file
gpp "binary search function" --create search.py

# Edit existing file
gpp "add error handling" --edit script.py

# Read and analyze
gpp "what does this do?" --read program.c

# Analyze without prompt
gpp --read code.java

Multiple File Operations

# Create multiple files
gpp "create a web app" --create index.html style.css app.js

# Edit multiple files with same instruction
gpp "add comments to all functions" --edit util.py helper.py main.py

# Analyze multiple files
gpp "how do these files interact?" --read app.py config.py database.py

Model Management

# List all available models
gpp --list-models

# Show currently loaded model
gpp --current-model

# Load a specific model
gpp --load-model "qwen2.5-coder-3b-instruct"

# Unload current model
gpp --unload-model

Advanced Options

# Specify language
gpp "quick sort algorithm" --lang=python --create sort.py

# Adjust generation parameters
gpp "add validation" --edit config.py --temperature=0.5 --max-tokens=800

📋 Supported Languages

Language	Extensions
C	.c, .h
C++	.cpp, .cc, .cxx, .hpp
Java	.java
Python	.py
HTML	.html, .htm
CSS	.css
PHP	.php
JavaScript	.js
TypeScript	.ts

📚 Examples

Example 1: Create a Web App

gpp "professional portfolio website with responsive design" \
    --create index.html styles.css script.js

Output:

1. index.html : [HTML code with semantic structure]
2. styles.css : [CSS code with responsive design]
3. script.js : [JavaScript for interactivity]

Example 2: Edit Multiple Files

gpp "add comprehensive docstrings and type hints" \
    --edit app.py utils.py models.py

Example 3: Analyze Project Structure

gpp "explain how these components work together" \
    --read main.py database.py api.py config.py

Example 4: Quick Code Generation

gpp "convert this algorithm to TypeScript" --lang=typescript
gpp "what are the security implications?" --read api.py

⚙️ Options Reference

Options:
  --create FILE(s)          Create one or multiple files
  --edit FILE(s)            Edit one or multiple files
  --read FILE(s)            Read and analyze files
  --lang, -l LANG           Specify programming language
  --temperature, -t FLOAT   Generation temperature (0.0-1.0, default: 0.7)
  --max-tokens, -m INT      Max tokens to generate (default: 500)

Model Management:
  --list-models             List all available models
  --current-model           Show currently loaded model
  --load-model MODEL        Load specific model
  --unload-model            Unload current model

  -h, --help                Show help message

🔧 Configuration

The tool uses these defaults:

• API: http://localhost:1234
• Model: qwen2.5-coder-3b-instruct
• Temperature: 0.7
• Max tokens: 500

To change defaults, edit the constants at the top of the gpp file:

API_BASE_URL = "http://localhost:1234"
MODEL_NAME = "qwen2.5-coder-3b-instruct"

🐛 Troubleshooting

"Cannot connect to LM Studio"

• Make sure LM Studio is running
• Verify it's accessible at http://localhost:1234
• Check your firewall settings

"Command not found: gpp"

• Run gpp --help from the installation directory
• Verify PATH is set correctly: echo $PATH
• Restart your terminal after installation

"Module not found: model_manager"

• Ensure model_manager.py is in the same directory as gpp
• Check file permissions: chmod +x model_manager.py

"requests library not found"

pip3 install requests

Model won't load

• Check if another model is already loaded
• Use gpp --unload-model first
• Verify sufficient disk space and RAM

📖 More Information

• Model Management: See MODEL_MANAGEMENT.md
• Implementation Details: See IMPLEMENTATION_SUMMARY.md
• Quick Reference: See MODEL_COMMANDS_QUICK_REFERENCE.txt

🤝 Contributing

To improve this tool:

1. Test with different models and edge cases
2. Report bugs and issues
3. Suggest new features

📝 License

This project is provided as-is for use with LM Studio.

🎯 Summary

GPP brings AI-powered code generation to your terminal with:

• ✅ Single and multiple file support
• ✅ Global terminal access from any directory
• ✅ Easy installation and setup
• ✅ Full model management capabilities
• ✅ Multiple language support
• ✅ Simple, intuitive command syntax

Get started now:

./install.sh
gpp --help