sheikh-cli 1.0.0

Privacy-first AI coding assistant for Termux

🧠 Sheikh-CLI

A privacy-first AI coding assistant that runs entirely on your Android device through Termux

Overview

The Termux Local Coding Agent is a powerful, privacy-focused development tool that transforms your Android device into a local AI coding assistant. Built specifically for Termux, it provides natural language interaction with your code while maintaining complete privacy through local operation.

✨ Features

🔒 Privacy-First

• 100% Local Operation - No cloud dependencies, no data transmission
• Secure Sandbox - Whitelisted commands with timeout protection
• Privacy Configuration - Full control over what data is processed

🛠️ Development Tools

• File Operations - Read, write, search, and manage files safely
• Code Analysis - Search patterns, find functions, analyze code structure
• Git Integration - Status, diff, commit, and branch management
• Shell Commands - Safe execution of development tools

🤖 AI-Powered

• Natural Language Interface - Describe tasks in plain English
• Intent Detection - Automatically maps requests to appropriate tools
• Local LLM Support - Integration with llama.cpp for offline inference
• No-Code CLI - Predefined actions for common development tasks

📱 Mobile-Optimized

• Termux Integration - Built specifically for Android Termux environment
• Resource Efficient - Optimized for mobile hardware constraints
• Battery Conscious - Intelligent resource management
• Portable Development - Full development environment in your pocket

🚀 Quick Start

Prerequisites

• Android device with Termux installed
• 4GB+ RAM (8GB recommended)
• 2GB+ free storage
• Internet connection for setup

Installation

1. Install Termux

   # Download from F-Droid or GitHub
   # Launch Termux and grant storage permissions
   termux-setup-storage
   

2. Run Setup Script

   # Copy setup_termux.sh to Termux
   bash setup_termux.sh
   

3. Start Using

   # Activate environment
   source ~/coding-agent/.venv/bin/activate
   
   # Interactive mode
   python ~/coding-agent/src/run_agent.py interactive
   
   # Single command
   python ~/coding-agent/src/run_agent.py prompt "list files in workspace"
   

💡 Usage Examples

File Operations

# Read a file
python src/run_agent.py prompt "read config.json"

# Create a new Python script
python src/run_agent.py prompt "create hello.py with print('Hello World')"

# List project files
python src/run_agent.py prompt "show all Python files in the project"

Code Analysis

# Search for patterns
python src/run_agent.py prompt "find all function definitions in Python files"

# Search imports
python src/run_agent.py prompt "show all import statements"

# Code statistics
python src/run_agent.py prompt "count lines of code in the project"

Git Operations

# Check repository status
python src/run_agent.py prompt "git status"

# View changes
python src/run_agent.py prompt "git diff"

# Show commit history
python src/run_agent.py prompt "git log --oneline -5"

Interactive Mode

# Start interactive session
python src/run_agent.py interactive

Example conversation:

You: help
Agent: [Shows available commands]

You: read the main script
Agent: [Reads and displays file content]

You: find all functions
Agent: [Searches and lists functions]

You: git status
Agent: [Shows git repository status]

You: quit
Agent: Goodbye! 👋

🏗️ Architecture

┌─────────────────┐    ┌──────────────────┐    ┌─────────────────┐
│   CLI Interface │    │  Orchestrator    │    │   Tool Plugins  │
│                 │    │                  │    │                 │
│ • Interactive   │───▶│ • Intent         │───▶│ • File Ops      │
│ • Single Prompt │    │   Detection      │    │ • Shell Cmds    │
│ • Help System   │    │ • Tool Routing   │    │ • Code Search   │
│                 │    │ • Error Handling │    │ • Git Ops       │
└─────────────────┘    └──────────────────┘    └─────────────────┘
         │                       │                       │
         │              ┌──────────────────┐              │
         └──────────────▶│   Security       │◀─────────────┘
                        │   Sandbox        │
                        │                  │
                        │ • Whitelisting   │
                        │ • Timeouts       │
                        │ • Directory      │
                        │   Restrictions   │
                        └──────────────────┘

🛡️ Security

Command Safety

• Whitelist Only - Only predefined safe commands execute
• Timeout Protection - Prevents infinite or long-running operations
• Directory Restrictions - Access limited to allowed directories
• Output Filtering - Sensitive data filtered from logs

Privacy Protection

• Local Processing - All AI inference happens on-device
• No Telemetry - Zero data transmission to external services
• Secure Storage - Configuration and logs encrypted when possible
• Audit Logging - Full activity tracking without sensitive data

⚙️ Configuration

Basic Configuration

{
    "model": {
        "type": "local_first",
        "llama_cpp_path": "/path/to/llama.cpp/main",
        "model_path": "/path/to/model.ggml"
    },
    "security": {
        "sandbox_enabled": true,
        "max_execution_time": 30,
        "allowed_directories": ["/home/coding-agent"]
    }
}

Advanced Options

• Model Integration - Support for various local LLMs
• Custom Tools - Add your own plugins
• Performance Tuning - Optimize for your device
• Extended Whitelist - Add trusted commands

📊 Performance

Recommended Hardware

Component	Minimum	Recommended
RAM	4GB	8GB+
Storage	2GB	5GB+
CPU	Quad-core	Octa-core

Optimization Tips

• Use smaller quantized models (4-bit)
• Enable battery optimization
• Monitor memory usage
• Clear logs regularly

🔧 Development

Project Structure

~/coding-agent/
├── src/
│   └── run_agent.py          # Main application
├── tools/
│   ├── file_operations.py    # File management
│   ├── shell_commands.py     # Safe shell execution
│   ├── code_search.py        # Code analysis
│   └── git_operations.py     # Git integration
├── config/
│   └── agent_config.json     # Configuration
├── models/                   # Local LLM models
├── logs/                     # Application logs
└── requirements.txt          # Python dependencies

Adding New Features

1. Create tool plugin in tools/
2. Register in orchestrator
3. Add intent detection
4. Update configuration
5. Document usage

🐛 Troubleshooting

Common Issues

Setup Fails

# Check Termux installation
pkg list-installed

# Update packages
pkg update && pkg upgrade

Permission Errors

# Grant storage access
termux-setup-storage

# Fix directory permissions
chmod -R 755 ~/coding-agent

Model Loading Issues

# Verify llama.cpp
cd ~/coding-agent/llama.cpp && ls -la main

# Test model
./main -m model.ggml -p "test"

Getting Help

1. Check the complete tutorial
2. Review troubleshooting section
3. Examine logs in ~/coding-agent/logs/
4. Test with simple commands first

🤝 Contributing

We welcome contributions! Please:

1. Follow the modular architecture
2. Add comprehensive error handling
3. Include security considerations
4. Document new features
5. Test on multiple devices

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• Termux Team - For the excellent Android terminal environment
• llama.cpp - For enabling local LLM inference
• Rich Library - For beautiful terminal interfaces
• Open Source Community - For inspiration and tools

📞 Support

• Documentation: See TUTORIAL.md
• Issues: Report bugs and feature requests
• Community: Join the Termux community

---

Built with ❤️ for developers who value privacy and portability

Transform your Android device into a powerful, private development environment with the Termux Local Coding Agent.