云效 DevOps MCP

阿里云云效 MCP 操作工具。通过 MCP SSE 协议操作云效平台，支持项目管理、工作项管理、流水线管理、代码仓库等 165+ 工具。

Security Scan

The skill's code, docs, and runtime instructions are consistent with a local MCP SSE client for 阿里云 云效 DevOps; it does not request unrelated credentials or contain surprising behavior, though deploying the companion MCP Server does require Alibaba Cloud credentials and must be done manually as documented.

Purpose & Capability

The name/description match the provided client code and SKILL.md: this is an MCP SSE client that talks to a local MCP Server (default http://localhost:3000) and exposes 165+ tools. There are no unrelated environment variables or binaries declared. The need for Alibaba Cloud credentials appears only in DEPLOY.md for the separate MCP Server image (reasonable and documented).

Instruction Scope

Runtime instructions only cover installing the Node client (npm install), starting/pointing to a local MCP Server, listing tools, querying schemas, and calling tools. The DEPLOY.md explicitly instructs that server deployment must be performed manually and that AI agents must not run the deployment commands. The SKILL.md and client code do not read arbitrary host files or exfiltrate data to external endpoints other than the configured MCP Server.

Install Mechanism

The registry entry has no automated install spec; the package includes a Node.js client whose installation is via npm (client recommends npm install). That pulls one dependency (eventsource-parser) from the public npm registry — expected for a Node client but worth noting as a standard third-party dependency fetch.

Credentials

The skill itself declares no required env vars or credentials. DEPLOY.md (separate MCP Server) requires ALIBABA_CLOUD_ACCESS_KEY_ID and ALIBABA_CLOUD_ACCESS_KEY_SECRET for server deployment; this is appropriate for a server that needs to access Alibaba Cloud, but these credentials are sensitive and the docs explicitly require manual, human-driven deployment rather than automated use by the agent.

Persistence & Privilege

No elevated privileges are requested. always is false, the skill does not request persistent insertion into agent configs, and it does not modify other skills or system-wide settings.

Assessment

This skill appears to be a local MCP SSE client and is internally consistent. Before installing or running it: - Do not hand your Alibaba Cloud RAM AccessKey/Secret to the agent; if you deploy the MCP Server (DEPLOY.md), follow the doc and perform deployment manually. The server Docker image will require those credentials. - Review and verify the Docker image source (build-steps-public-registry.cn-beijing...) before running it. - npm install will fetch a public dependency (eventsource-parser); run installs in a controlled environment and review dependencies if you have strict supply-chain requirements. - The client talks to a user-configurable baseUrl (defaults to localhost:3000). Ensure you only point it at a server you trust. If you want higher assurance, ask for: the publisher identity, a signed release or official upstream repo, and an audit of the Docker image used by the MCP Server.

云效 MCP Skill

阿里云云效（DevOps）MCP 操作工具，基于 MCP SSE 协议。MCP Server: http://localhost:3000

目录结构

yunxiao-mcp/
├── SKILL.md           # 本文件
├── DEPLOY.md          # MCP Server 部署与运维
└── client/            # Node.js MCP 客户端
    ├── package.json
    ├── examples/
    └── src/
        ├── client.mjs # MCP SSE 客户端核心
        ├── cli.mjs    # 命令行接口
        └── index.mjs  # 模块导出

功能概览

通过 MCP Server 提供 165+ 工具，覆盖：

• 项目管理: 项目查询、迭代管理、版本管理
• 工作项管理: 需求、任务、缺陷的查询与操作
• 流水线管理: 流水线创建、运行、日志查询
• 代码管理: 仓库、分支、文件操作
• 部署管理: 部署单、环境管理

快速开始

# 安装客户端
cd client && npm install

# 确保 MCP Server 运行（未运行参考 DEPLOY.md）
docker ps --filter "name=yunxiao-mcp"

命令速查

# 在项目 client 目录下执行以下命令

# 查看所有工具（Markdown 表格，省 token）
node src/cli.mjs list

# 查看某个工具的参数定义（JSON，含 inputSchema）
node src/cli.mjs schema get_current_user

# 当前用户
node src/cli.mjs call get_current_user

# 当前企业信息
node src/cli.mjs call get_current_organization_info

# 搜索项目
node src/cli.mjs call search_projects '{"organizationId": "<org_id>", "perPage": 20}'

# 搜索需求
node src/cli.mjs call search_workitems '{"organizationId": "<org_id>", "spaceId": "<project_id>", "category": "Req", "perPage": 20}'

# 搜索 Bug
node src/cli.mjs call search_workitems '{"organizationId": "<org_id>", "spaceId": "<project_id>", "category": "Bug", "perPage": 20}'

# 获取版本列表
node src/cli.mjs call list_versions '{"organizationId": "<org_id>", "id": "<project_id>", "perPage": 20}'

# 从 stdin 传递大参数
echo '{"organizationId": "<org_id>"}' | node src/cli.mjs call search_projects --stdin

典型 AI agent 调用流程

# 1. 浏览工具列表（Markdown 表格，低 token）
node src/cli.mjs list

# 2. 获取目标工具参数定义（JSON，按需获取）
node src/cli.mjs schema search_workitems

# 3. 执行调用（JSON 结果）
node src/cli.mjs call search_workitems '{"organizationId":"xxx","spaceId":"yyy","category":"Req"}'

常用工具速查

项目管理

工具名	说明
search_projects	搜索项目列表
get_project	获取项目详情
list_versions	获取版本列表
list_sprints	获取迭代列表

工作项管理

工具名	说明
search_workitems	搜索工作项（需求/Bug/任务）
get_work_item	获取工作项详情
create_work_item	创建工作项
update_work_item	更新工作项

流水线管理

工具名	说明
list_pipelines	获取流水线列表
get_pipeline	获取流水线详情
create_pipeline_run	运行流水线
get_pipeline_run	获取运行记录

代码管理

工具名	说明
list_repositories	获取代码仓库列表
list_branches	获取分支列表
get_file_blobs	获取文件内容
create_file	创建文件

参数说明

必填参数

参数	说明
organizationId	企业 ID，可通过 get_current_organization_info 获取
spaceId	项目 ID
category	工作项类型，取值见下表

工作项类型 (category)

值	说明
Req	需求
Bug	缺陷
Task	任务
Risk	风险
Topic	主题
Request	诉求

Node.js API 使用

import { createClient } from './src/index.mjs';

const client = await createClient('http://localhost:3000');
const result = await client.callTool('search_workitems', {
  organizationId: '<org_id>',
  spaceId: '<project_id>',
  category: 'Req',
  perPage: 10,
});
console.log(JSON.stringify(result));
await client.close();

相关资源

• MCP Server 部署与运维: DEPLOY.md
• 云效官方文档: https://help.aliyun.com/zh/yunxiao/
• MCP 协议: https://modelcontextprotocol.io/