Tangku AgentOS

Tangku AgentOS is a modular Agent Operating System for autonomous AI agents, workflows, and runtime orchestration. It combines kernel-based runtime supervision with extensible engines, provider integrations, and a production-ready dashboard shell.

Current Status

• Version: v1.0.0-beta RC2
• Completion: ~ongoing
• Regression Tests: Passing
• Release Candidate: Verified for broader production readiness and packaging consistency

Project Overview

Tangku AgentOS provides an opinionated runtime platform for building agent-based automation across planning, execution, workflow, tools, and observability. It is designed to make AI agents composable, provider-agnostic, and workspace-aware without requiring a single monolithic application.

Architecture

The repository is structured around a central kernel and specialized runtime subsystems:

• Kernel Runtime: coordinates startup, lifecycle events, and runtime supervision
• Feature Runtimes: planning, workflow, execution, tool runtime, automation, browser, plugin, and multi-agent runtimes
• Supporting Engines: memory, knowledge, workspace, repository, security, and observability
• Provider Layer: provider-agnostic integration for AI backends and model providers
• Interface Layer: web dashboard shell and HTTP adapter for runtime status and control

Key Features

• Modular runtime architecture with clear separation of concerns
• Provider abstraction for AI model and service integrations with routing, fallback, and health awareness
• Workflow, automation, and scheduler support for parallel and dependency-driven execution
• Multi-agent coordination runtime for discovery, delegation, shared memory, and recovery
• Memory and knowledge engines for stateful agent behavior and versioned coordination
• Plugin and extension runtime scaffolding with runtime lifecycle hooks
• Browser automation and terminal execution support
• Observability, logging, and health/status reporting
• Production web dashboard shell for runtime state, coordination insights, and command palette display
• Python packaging with wheel and source distribution support

Repository Structure

• tangku_agentos/ — main package with runtimes, engines, and integration components
• docs/ — documentation and release notes
• tests/ — regression and smoke tests
• benchmarks/ — benchmarks and performance placeholders
• examples/ — example usage content and scenarios
• scripts/ — development helper scripts
• LICENSE / LICENSE.txt — license terms
• VERSION / VERSION.txt — project version metadata
• .github/ — contribution and issue templates

Installation

git clone https://github.com/gauryat/TangkuAgentOS.git
cd TangkuAgentOS
python -m pip install -e .

Quick Start

python -m pip install -e .
pytest -q
python -m build --sdist --wheel

To launch the dashboard server:

python -m tangku_agentos.interface_layer.web_dashboard_server

Development Status

Tangku AgentOS is nearing release readiness with the core runtime packages implemented and validated. Current work is focused on integration stabilization, documentation, packaging, and release candidate verification.

Roadmap

Planned milestones include:

• provider adapter backends and richer model integrations
• repository-backed automation and Git-enabled workflows
• browser automation and terminal orchestration
• plugin ecosystem and runtime extensions
• vector database and external knowledge persistence
• production stability, performance, and observability enhancements

Contribution Guide

Please read CONTRIBUTING.md before contributing. Follow these steps:

1. Open an issue for new features or bug reports.
2. Create a topic branch for your change.
3. Add tests and update documentation.
4. Submit a pull request using the project PR template.

License Summary

Tangku AgentOS is released under the MIT License. See LICENSE or LICENSE.txt for full terms.

Repository Links

• Repository: https://github.com/gauryat/TangkuAgentOS
• Issues: https://github.com/gauryat/TangkuAgentOS/issues
• Pull requests: https://github.com/gauryat/TangkuAgentOS/pulls
• Security: https://github.com/gauryat/TangkuAgentOS/security/advisories