open-agentkit

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

★ 22 Open GitHub ↗

Open-source alternative to OpenAI AgentKit — build, manage, and deploy AI agents with full transparency and flexibility.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# OAK: Open AgentKit

Inspired by OpenAI AgentKit, we believe the community equally needs a truly open and universal "AgentKit"—one that supports all models and ecosystem tools, integrates industry best practices, and adheres to mainstream Agent protocols and standards.

## Why OAK
- **Practical origin, open execution**: OpenAI AgentKit offers useful ideas and components, but it is closed-source and tightly coupled to a single ecosystem, limiting flexibility across many real-world scenarios.
- **Open and universal**: The community needs a vendor-neutral AgentKit that integrates industry best practices (e.g., Claude Skills), supports mainstream Agent protocols and standards, and works across models, tools, and platforms.

## What is OAK
OAK (Open AgentKit) is an open-source initiative to build a general-purpose, ecosystem-agnostic AI Agent platform together with global developers.

Our north star is to be a fully open-source counterpart to OpenAI AgentKit while remaining model- and platform-agnostic.

## Core Modules
- **OAK Framework ([LangCrew](https://github.com/01-ai/langcrew))**: A full-stack multi-Agent development framework enabling efficient architecture and collaboration.
- **OAK Builder**: A community-built visual Agent composer that outputs production-ready code.
- **OAK Runtime**: A stable, reliable runtime for executing and orchestrating Agents at scale.
- **OAK Studio**: A visual workbench for deployment, evaluation, tracing, and continuous optimization.

## Roadmap (High-level)
- [x] Multi-Agent Framework (LangCrew): A full-stack development framework.
- [ ] Ship Builder for visual composition and code generation.
- [ ] Provide evaluation toolchain: datasets, tracing, scoring, and automated prompt optimization.
- [ ] Release embeddable chat and Agent UI components.
- [ ] Stabilize Runtime for reliable, observable multi-Agent execution.
- [ ] Support Claude Skills integration to enhance Agent capabilities with Claude's skill ecosystem.
- [ ] Expand standards compliance and ecosystem integrations.

## Contributing
We welcome issues and pull requests from the community. Please follow the repository’s coding style and contribution guidelines. Together with OSChina and the broader open-source community, we aim to build an open, collaborative, and sustainable Agent ecosystem.

## License
This project is licensed under the MIT License. See `LICENSE` for details.

## Stay in the loop
Star and watch `01-ai/open-agentkit` to follow OAK’s latest updates. Join us in building the open-source Agent infrastructure and adapters the community deserves.

open-agentkit

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README