OpenSIN — The World's Most Comprehensive AI Agent System

[!IMPORTANT]
SSOT: Die kanonische OpenCode-Konfiguration liegt unter Delhi/upgraded-opencode-stack.
Nach jeder Änderung MUSS sin-sync ausgeführt werden.

[!IMPORTANT]
SSOT: Die kanonische OpenCode-Konfiguration liegt unter Delqhi/upgraded-opencode-stack.
Nach jeder Änderung MUSS sin-sync ausgeführt werden.

OpenSIN is a comprehensive AI agent system that enables you to build, deploy, and manage AI agents at scale. Built from the ground up for multi-agent collaboration with our A2A (Agent-to-Agent) protocol.

🚀 Quick Start

# Install the CLI
pip install opensin-cli

# Create your first agent
opensin agent create researcher --model gpt-4

# Test your agent
opensin agent test researcher --prompt "What is AI?"

# Create a team
opensin team create research-team --strategy sequential

# Execute a team task
opensin team exec research-team --task "Research AI trends"

📦 Packages

Package	Description	Status
opensin-core	Core engine with QueryEngine, Hook System, Tool System, Permission System, Subagent System, MCP Client, Sandbox, Memory System	✅ Stable
opensin-cli	Command-line interface for managing agents and teams	✅ Stable
opensin-api	FastAPI REST API server	✅ Stable
opensin-sdk	Python SDK for programmatic access	✅ Stable

🧠 Core Features

QueryEngine

Async generator pattern with ReAct loop, token budget tracking, and automatic context compaction.

Hook System

20+ hook events with 5 execution modes (shell, HTTP, function, agent, prompt) for powerful automation.

Tool System

Rich tool interface with 8 built-in tools, deferred loading, and permission-aware filtering.

Permission System

Multi-source permission rules with auto-mode classifier and bypass-immune safety checks.

Subagent System

Fork pattern with cache-identical prefixes, worktree isolation, and background-to-foreground transitions.

MCP Client

Multi-transport support (stdio, SSE, HTTP, WS, in-process) with OAuth authentication.

Sandbox

Secure code execution with filesystem/network rules and git escape prevention.

Memory System

File-based memory with index and topic files, SIN.md auto-discovery, and session memory extraction.

📚 Documentation

Full documentation is available at docs.opensin.ai, with 323 pages covering:

🐳 Local Development

# Clone the repository
git clone https://github.com/OpenSIN-AI/OpenSIN.git
cd OpenSIN

# Setup development environment
./scripts/setup.sh

# Start all services
docker-compose up -d

# Run tests
./scripts/test.sh

🧪 Testing

# Run all tests
./scripts/test.sh

# Run specific test suites
pytest opensin_core/tests/ -v
pytest opensin_cli/tests/ -v
pytest opensin_api/tests/ -v
pytest opensin_sdk/tests/ -v
pytest tests/integration/ -v

🤝 Contributing

We welcome contributions! Please see our Contributing Guide for details.

Fork the repository
Create your feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request

📊 Project Stats

323 documentation pages
15,934+ lines of code
657 tests passing
11 core modules
4 packages
153 repositories with docs links

🌐 Community

Documentation: https://docs.opensin.ai
GitHub: https://github.com/OpenSIN-AI
Discord: https://discord.gg/opensin
Issues: https://github.com/OpenSIN-AI/OpenSIN/issues

📄 License

This project is licensed under the Apache 2.0 License - see the LICENSE file for details.

Built with ❤️ by the OpenSIN team