AI Skill Hub 强烈推荐:协同AI工作流 是一款优质的Agent工作流。AI 综合评分 8.0 分,在同类工具中表现稳健。如果你正在寻找可靠的Agent工作流解决方案,这是一个值得深入了解的选择。
协同AI工作流 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
协同AI工作流 是一套完整的 AI Agent 自动化工作流方案。通过可视化的节点编排,将复杂的多步骤任务拆解为清晰的自动化流程,实现全程无人值守的智能处理。支持与数百种外部服务和 API 无缝集成,适合构建数据处理管线、业务自动化和 AI 辅助决策系统。
# 方式一:go install(推荐) go install github.com/AndrewDryga/coop@latest # 方式二:从源码编译 git clone https://github.com/AndrewDryga/coop cd coop go build -o coop . # 方式三:下载预编译二进制 # 访问 Releases 页面下载对应平台二进制文件 # https://github.com/AndrewDryga/coop/releases
# 查看帮助 coop --help # 基本运行 coop [options] <input> # 详细使用说明请查阅文档 # https://github.com/AndrewDryga/coop
# coop 配置说明 # 查看配置选项 coop --config-example > config.yml # 常见配置项 # output_dir: ./output # log_level: info # workers: 4 # 环境变量(覆盖配置文件) export COOP_CONFIG="/path/to/config.yml"
curl -fsSL https://coop.dryga.com/install.sh | sh
The script downloads the coop binary for your system into ~/.local/bin. It needs curl, tar, and sha256sum or shasum. If a container runtime is available, it also builds the box image and runs coop doctor. If not, install one and then run:
coop build --egress open && coop doctor
co:op runs on macOS and Linux with Docker. Online provider runs use Docker's private network namespace to keep credentials outside the box. Apple's container remains available for raw open-network workloads, not provider-brokered or filtered runs. co:op prefers Docker when both are present. The coop binary is static.
coop update updates the binary, then uses the installed version to prepare the image for this project's network mode. Filtered projects build on the pinned client base; open and offline images refresh their base components. If a filtered project needs a build later, an interactive provider launch can offer Build and continue. Review the saved snapshot before approving; unattended runs still require an explicit host build. Each release carries pinned agent CLIs and ACP adapters. To build from source, run git clone https://github.com/AndrewDryga/coop && cd coop && make install.
<a name="verifying-a-download"></a> <details><summary>Verifying a download</summary>
The one-line command runs install.sh from the main branch, served by GitHub Pages at coop.dryga.com. That trusts GitHub, the domain, and the current main branch. To trust GitHub alone, fetch the same script from https://raw.githubusercontent.com/AndrewDryga/coop/main/install.sh.
The script then needs the release's checksums.txt, and checks the downloaded archive against it with sha256sum or shasum. It stops if the checksum file or the tool is missing. When cosign is installed, the script also proves the checksum file came from the release workflow for the exact tag you asked for. Every release since v2.2.2 ships that signature bundle, so a missing bundle stops the install instead of falling back to the checksum alone. Without cosign, the script says the signature wasn't checked.
To verify before you run any project code, download the release files by hand. Set VER and ASSET for your platform, for example VER=v8.1.0 ASSET=coop_8.1.0_darwin_arm64.tar.gz:
```bash base="https://github.com/AndrewDryga/coop/releases/download/$VER" curl -fsSLO "$base/$ASSET" curl -fsSLO "$base/checksums.txt" curl -fsSLO "$base/checksums.txt.bundle"
co:op is one static Go binary. make install builds it into ~/.local/bin, and make check runs the same gate CI runs. CONTRIBUTING.md covers the source layout and the test layers.
coop login claude # 1. sign in, once per account
cd ~/code/your-repo # 2. any git repo
coop init # 3. set up the project: settings, task queue, hooks
coop claude # 4. start the agent in the box
coop init sets up the task queue, hooks, shared instructions and skills, and the selected agent directories. It configures the agents you're signed in to, so sign in first.
Each agent starts with its own "don't stop to ask" flag on (--dangerously-skip-permissions, --dangerously-bypass-approvals-and-sandbox, --yolo). An agent that goes off the rails works on the repo you gave it, which you can restore from Git, and the rest of your computer stays out of reach. Anything after the agent's name goes to the agent, so coop claude --continue resumes Claude's last session in the box. Codex is the exception: its -p picks a profile, so run a one-shot prompt with coop codex exec "…".
Other ways in:
coop codex # the same box, with Codex
coop claude --peer codex --peer gemini # read-only second opinions for hard calls
coop shell # a shell in the box, to look around
coop run -- npm test # any command in the box
cosign verify-blob checksums.txt \ --bundle checksums.txt.bundle \ --certificate-identity "https://github.com/AndrewDryga/coop/.github/workflows/release.yml@refs/tags/$VER" \ --certificate-oidc-issuer https://token.actions.githubusercontent.com
coop 是一个专为开发者设计的沙盒化 AI Agent 运行环境。它允许你在隔离的容器中安全地运行 Claude、Codex 或 Gemini 等 AI 模型,确保 AI 在处理本地代码库时既能获得完整的上下文,又不会泄露敏感信息或破坏系统环境。
使用 coop 需要系统中已安装容器运行时(Container Runtime),支持 Docker、Podman 或 macOS 26+ 自带的 Apple container。此外,运行特定的 AI 模型需要相应的 API 凭证。
可以通过官方提供的安装脚本一键完成部署:使用 `curl -fsSL https://raw.githubusercontent.com/AndrewDryga/coop/main/install.sh | sh` 将预编译的 `coop` 二进制文件下载至 `~/.local/bin`。安装完成后,若系统已存在容器运行时,脚本会自动构建沙盒镜像并运行 `coop doctor` 进行环境检查;若未检测到运行时,请手动执行 `coop build && coop doctor`。
coop 的使用非常直观。首先进入你的代码仓库目录,通过 `coop login claude` 进行一次性身份验证。随后,你可以直接运行 `coop claude`、`coop codex` 或 `coop gemini` 来启动对应的沙盒 Agent。此外,`coop fusion` 支持多模型协作模式,而 `coop shell` 和 `coop run` 则允许你在沙盒环境中执行命令或测试脚本,实现完全隔离的开发体验。
coop 采用“一箱三 Agent”的设计理念。每个 Agent 的配置与凭证均存储在宿主机的 `~/.config/coop/agents/<name>/` 目录下,并以挂载方式映射到沙盒内的对应路径(如 `~/.claude`)。这种设计确保了敏感凭证不会进入 Git 仓库。此外,coop 支持通过 `.tool-versions` (asdf) 实现零配置环境,能够根据仓库定义的版本自动在运行时配置工具链。
coop 的所有命令均基于当前工作目录的仓库运行。核心命令包括:`coop [agent]` 用于启动沙盒化的 AI Agent;`coop fusion [agent]` 用于启动多模型治理模式;`coop fork` 则允许你将特定的分支交给 Agent 在临时克隆环境中处理,实现安全的代码重构与评审。
项目的发布流程高度自动化且安全,通过 GitHub Release Workflow 进行签名。利用 cosign 技术对 `checksums.txt` 进行校验,确保分发的二进制文件在传输过程中未被篡改,符合现代云原生安全标准。
针对常见问题,若遇到 "no container runtime found" 错误,请安装 Docker 或 Podman,或通过 `COOP_RUNTIME=docker` 强制指定运行时。若提示镜像未构建,请执行 `coop build`。所有安装包均通过 GitHub Workflow 进行签名,确保安全性。
高质量的开源AI工作流项目,具有自动化和安全性
AI Skill Hub 为第三方内容聚合平台,本页面信息基于公开数据整理,不对工具功能和质量作任何法律背书。
建议在沙箱或测试环境中充分验证后,再部署至生产环境,并做好必要的安全评估。
✅ MIT 协议 — 最宽松的开源协议之一,可自由商用、修改、分发,仅需保留版权声明。
总体来看,协同AI工作流 是一款质量优秀的Agent工作流,在同类工具中具备一定竞争力。AI Skill Hub 将持续追踪其更新动态,建议收藏备用,结合自身场景选择合适时机引入使用。
| 原始名称 | coop |
| 原始描述 | 开源AI工作流:Run a coding agent on your real repos every day — in a box it can't escape, with。⭐18 · Go |
| Topics | aillmsandboxgo |
| GitHub | https://github.com/AndrewDryga/coop |
| License | MIT |
| 语言 | Go |
收录时间:2026-06-24 · 更新时间:2026-06-26 · License:MIT · AI Skill Hub 不对第三方内容的准确性作法律背书。
选择 Agent 类型,复制安装指令后粘贴到对应客户端