JiuwenSwarm
Understands Your Intent, Evolves Autonomously — Swarm Collaboration for Complex Tasks
Chinese
·
English
·
Docs
·
Docs (EN)
·
Website
·
GitCode
---
## Introduction
**JiuwenSwarm** is an Agent system that makes multi-agent collaboration truly work. Designed for developers and teams who need to automate complex tasks, it helps users drive multi-agent collaboration, Skill self-evolution, and tool invocation through natural language — delivering end-to-end from intent to result. Compared to similar projects, JiuwenSwarm's core differentiators are: Skill self-evolution makes capabilities grow stronger with use, swarm collaboration enables multiple Agents to specialize and coordinate on complex tasks, and multi-platform access covers mainstream IM platforms.
### Why JiuwenSwarm
| Capability | Value |
| --- | --- |
| Multi-Agent Collaboration | Complex tasks are no longer a single Agent's battlefield: Leader automatically decomposes tasks, assembles teams, and multiple Agents specialize and negotiate dynamically |
| Distributed Agent Swarm | Break through single-machine compute limits: Leader/Teammate deploy across processes and machines, coordinating at scale |
| Swarmflow | Orchestrate workflows with natural language: Leader automatically decomposes into multi-stage workflows, with Agents seamlessly handing off between stages. Human-in-the-loop support allows you to answer questions mid-workflow via `/swarmflow on` and the TUI P key interface |
| Skill Self-Evolution | Capabilities grow stronger with use, not rigid over time: automatically detects error signals and user dissatisfaction, then optimizes Skill definitions |
| Skill Hub Sharing | Capability assets are built once and reused everywhere: Skills can be searched, installed, combined, and remixed across developers |
| Auto Harness | Evaluation-driven, end-to-end automated Harness optimization: not training model weights, but letting the Harness learn and optimize automatically in practice |
| AI Infrastructure Compatibility | One system adapts to multiple inference backends: compatible with Huawei Cloud MaaS and other mainstream platforms, supports OpenAI-compatible APIs and local models |
| Tool Permissions & Security | Every step is under your control: pre-execution approval for tools, file access whitelists, sensitive operation interception |
## Latest Updates
- **2026-05-18**: `v0.2.0` released — JiuwenClaw officially upgraded to JiuwenSwarm, desktop app with auto-update, Swarm mode and evolution enhancements, CLI/TUI command extensions.
- **2026-05-21**: openJiuwen Roundtable livestream — open sharing on Swarm Skills Hub ecosystem co-building and Skills self-evolution.
- **2026-06-02**: `v0.2.1` released — Swarm cluster mode frontend display optimization, TUI integration with Auto Harness new commands, automated evolution and scheduled task system improvements.
- **2026-06-12**: Invited to Singapore Lorong AI community event.
## Installation & Launch
### Desktop
| Platform | Download | Notes |
| --- | --- | --- |
| Windows | [Download Windows Version](https://openjiuwen.com/en/jiuwenswarm) | For Windows 10 / 11 |
| macOS | [Download macOS Version](https://openjiuwen.com/en/jiuwenswarm) | For Intel / Apple Silicon |
Download and follow the installer prompts to get started.
### Command Line
```bash
# Install JiuwenSwarm
pip install jiuwenswarm
# Use China mirror (recommended)
pip install jiuwenswarm -i https://pypi.tuna.tsinghua.edu.cn/simple
# Initialize JiuwenSwarm (first-time setup)
jiuwenswarm-init
# Start JiuwenSwarm
jiuwenswarm-start
```
After launching, visit http://localhost:5173 to open the frontend.
To use TUI (terminal interface), open a new terminal after starting JiuwenSwarm:
```bash
# Install JiuwenSwarm-tui
pip install jiuwenswarm-tui
# Use China mirror (recommended)
pip install jiuwenswarm-tui -i https://pypi.tuna.tsinghua.edu.cn/simple
# Start JiuwenSwarm-tui
jiuwenswarm-tui
```
> For detailed installation instructions, see: [Install Guide](docs/en/InstallGuide.md)
## Quick Start
### Configure Model
JiuwenSwarm supports multiple model platforms: Huawei Cloud MaaS, OpenAI, DeepSeek, DashScope, SiliconFlow, OpenRouter and other OpenAI-compatible APIs, as well as local model deployment.
### Start a Conversation
JiuwenSwarm supports three execution modes — switch as needed:
| Mode | Description | Best For |
| --- | --- | --- |
| Plan Mode | Decomposes requirements into concrete steps, executes step by step | Complex tasks, need to confirm each step |
| Performance Mode | Flexible handling, supports parallel tasks | Simple tasks, fast response |
| Swarm Mode | Leader orchestrates multiple specialized Agents in coordinated collaboration | Large complex tasks, multi-role collaboration |
**Swarm Mode** (default)
Example input:
```text
Conduct an in-depth research on the new energy vehicle industry and generate an analysis report.
```
**Planning Mode**
Example input:
```text
Help me build a personal blog website — from tech selection, page design, frontend and backend development to deployment, create a complete development plan and execute step by step.
```
**Performance Mode**
Example input:
```text
Check today's weather in Beijing, and recommend 3 books about artificial intelligence.
```
### TUI Interface
Experience JiuwenSwarm in the terminal — ideal for headless environments or CLI-preferred users.
Example input:
```text
Write a bubble sort algorithm for me.
```
> For detailed operation guide, see: [Quick Start](docs/en/Quickstart.md)
## Documentation
For common usage instructions and feature documentation, see: [Documentation](docs/README_EN.md)
## Roadmap
| Feature | Status | ETA | Value |
| --- |-----| --- | --- |
| Swarmflow Stateful Operators | Shipped | 2026-07 | Workflow nodes support human intervention and state memory — Humans can participate as stateful operators, approving, correcting, or taking over subsequent steps based on intermediate results |
| Team Mode Same-Session Feishu Integration | In Development | 2026-07 | Feishu users join as Human Agents in the same session, collaborating with AI Agent teams — human-AI mixed squads for more precise decisions |
## FAQ
For solutions to common issues, see: [FAQ](docs/en/FAQ.md).
## Contributing
We welcome developers to contribute to JiuwenSwarm. You can contribute in the following ways:
- Report bugs, feature requests, or usage issues: [Issues](https://github.com/openJiuwen-ai/jiuwenswarm/issues)
- Submit code, documentation, or examples: [Pull Requests](https://github.com/openJiuwen-ai/jiuwenswarm/pulls)
- Share Skills: [Swarm Skills Hub](https://swarmskills.openjiuwen.com/)
Please read the [Contributing Guide](docs/en/Contributing.md) before contributing to understand the debugging workflow, code style, and commit conventions. For the contribution roadmap, see the [openJiuwen Contribution Page](https://openjiuwen.com/en/contribute).
### Contributors
Thanks to all developers who have contributed to JiuwenSwarm: [View Contributor List](https://github.com/openJiuwen-ai/jiuwenswarm/graphs/contributors)
## Community
| Channel | Purpose | Link |
| --- | --- | --- |
| Website | Product info, updates, and ecosystem | [Visit Website](https://openjiuwen.com/en/) |
| SIG | Technical roadmap, engineering practices, ecosystem building | [Join SIG](https://openjiuwen.com/en/community/sig-center) |
| Swarm Skills Hub | Browse, publish, and reuse JiuwenSwarm Skills | [Visit Swarm Skills Hub](https://swarmskills.openjiuwen.com/) |
## License
This project is licensed under [Apache License 2.0](LICENSE).
This product serves solely as a workflow orchestration tool and does not embed any AI model capabilities. When users integrate AI models for specific business scenarios, they shall bear full responsibility for compliance obligations under the EU AI Act and other relevant regulatory frameworks.