openclaudecode 0.1.33

Claude Code - An AI-powered code assistant with comprehensive tools

Claude Code

An AI-powered development assistant with comprehensive tools for code analysis, file operations, web research, and task management. Built with Ollama integration and inspired by Anthropic's Claude Code.

Features

🛠️ Comprehensive Tool Suite

• File Operations: Read, write, edit files with advanced string replacement
• Search & Discovery: Glob patterns, regex search, directory listing
• Code Execution: Bash command execution with safety checks
• Jupyter Support: Read and edit notebook cells
• Web Integration: Fetch and analyze web content
• Task Management: Built-in todo system for complex workflows

🤖 AI Agent Modes

• Autonomous Mode: Agent works step-by-step with full tool access
• Interactive Mode: Manual control over each tool execution
• Chat Mode: Direct conversation without tools
• Streaming Mode: Real-time response streaming

🎨 Rich Terminal UI

• Claude-inspired color scheme and branding
• Syntax highlighting for code blocks
• Interactive file and command completion
• Structured output with panels and tables
• Live context display

🔧 Model Integration

• Default Model: qwen2.5:7b-instruct (efficient and capable)
• Ollama Backend: Local model hosting for privacy
• Configurable Context: Adjustable context window sizes
• Tool Schema: Comprehensive function calling support

Installation

Prerequisites

• Python 3.11+
• Ollama server running locally or remotely
• qwen2.5:7b-instruct model (or configure alternative)

Install from Source

git clone https://github.com/openclaudecode/openclaudecode.git
cd openclaudecode
pip install -e .

Pull Required Model

ollama pull qwen2.5:7b-instruct

Quick Start

Basic Usage

# Start Claude Code in current directory
claude

# Specify workspace and model
claude --workspace /path/to/project --model qwen2.5:7b-instruct

# Run with custom Ollama host
claude --host http://192.168.1.100:11434

# Execute single query
claude --query "analyze the project structure"

Commands

Command	Description
/help	Show available commands
/read <file>	Read file with syntax highlighting
/write <file>	Write content to file
/edit <file>	Interactive file editing
/ls <path>	List directory contents
/grep <pattern>	Search files with regex
/bash <command>	Execute shell command
/todo	View/manage todo list
/chat <message>	Chat without tools
/stream <message>	Stream response
/interactive <message>	Interactive mode
@filename	Add file to context

Example Workflow

# Start Claude Code
claude

# Add files to context
@src/main.py
@tests/test_main.py

# Analyze and improve code
> analyze the code structure and suggest improvements

# Execute suggested changes
> implement the logging improvements you suggested

# Run tests
/bash python -m pytest tests/

# Create documentation
> create comprehensive README.md for this project

Configuration

Model Configuration

# List available models
ollama list

# Use different model
claude --model llama2:7b

# Adjust context size
claude --num-ctx 8192

Host Configuration

# Local Ollama (default)
claude --host http://localhost:11434

# Remote Ollama
claude --host http://192.168.1.100:11434

# Custom port
claude --host http://localhost:8080

Tool Reference

File Operations

• read: Read files with line numbers and syntax detection
• write: Create or overwrite files completely
• edit: Precise string replacement with conflict detection
• multiedit: Atomic multiple edits in single operation

Search & Discovery

• glob: Fast pattern-based file finding (*.py, **/*.js)
• grep: Content search with regex support
• ls: Directory listing with ignore patterns

Code Execution

• bash: Safe command execution with timeout and output limits

Specialized

• notebook_read/edit: Jupyter notebook cell manipulation
• web_fetch: Web content retrieval and analysis
• todo_read/write: Session task management
• task: Autonomous sub-agent deployment

Security

Command Safety

• Dangerous command patterns blocked
• Execution timeouts prevent hanging
• Output size limits prevent overflow
• Workspace isolation for file operations

Privacy

• Local model execution (no external API calls)
• Workspace-scoped file access
• No data transmission to external services

Development

Project Structure

claude_code/
├── src/
│   ├── app.py          # Main CLI application
│   ├── agent.py        # AI agent orchestrator
│   ├── llm.py          # Ollama client
│   └── tools.py        # Tool implementations
├── cli_entry.py        # CLI entry point
└── __init__.py

Testing

# Install development dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run specific test
pytest tests/test_tools.py::test_read_file

Contributing

1. Fork the repository
2. Create feature branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'Add amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open Pull Request

Troubleshooting

Common Issues

Model not found

# Pull the required model
ollama pull qwen2.5:7b-instruct

# Or use different model
claude --model llama2:7b

Connection refused

# Start Ollama service
ollama serve

# Check if running
curl http://localhost:11434/api/tags

Permission errors

# Ensure proper workspace permissions
chmod -R 755 /path/to/workspace

Performance Tuning

Memory optimization

# Reduce context size for lower memory usage
claude --num-ctx 2048

Response speed

# Use smaller, faster model
claude --model qwen2.5:1.5b

License

MIT License - see LICENSE file for details.

Acknowledgments

• Inspired by Anthropic's Claude Code
• Built with Ollama for local AI execution
• Powered by qwen2.5 model family
• Rich terminal UI with Python Rich library