workbench-ecs
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
workbench-ecs
把 dsh-workbench-ecs(DeepSeek Harness 的阿里云 ECS 插件,v0.9.0)移植为 WorkBuddy 的 MCP 服务器。
本仓库路径:
E:\AiProject\workbench-ecs(唯一有效来源)移植来源:
C:\Users\ASUS\.dsh\node_modules\dsh-workbench-ecs(github.com/nishuoyang/dsh-workbench-ecs)运行方式: 零依赖 Node.js(≥18)MCP stdio 服务器,
src/server.mjs单文件底层通道: 复用同一本机 阿里云 Workbench CLI(
~/.workbench/config.json凭据),实例经 Workbench 后端连接,无需公网 IP / SSH 密钥
目录结构
workbench-ecs/
├── src/
│ └── server.mjs # MCP 服务器入口(唯一代码文件,零依赖)
├── scripts/
│ └── smoke.mjs # 冒烟测试:initialize + tools/list 握手校验
├── config/
│ └── mcp.example.json # 连接器配置片段(粘贴进 ~/.workbuddy/mcp.json)
├── docs/
│ └── MIGRATION.md # 迁移记录(来源、变更、验证)
├── package.json # 依赖声明(无运行时依赖)+ scripts
├── .gitignore
└── README.mdRelated MCP server: RunbookBridge
提供的工具(8 个)
工具 | 说明 |
| 列出指定地域的 ECS 实例 |
| 跨地域关键字检索(缺省扫全部 22 个公共地域) |
| 远程执行命令/脚本(script 走 base64 直送零转义) |
| 按字节游标续读远端日志文件 |
| 上传文件/目录(sha256 校验、瞬时失败重试、目录 tar 自动解包) |
| 下载远端文件到本机 |
| 7 段预置只读体检(主机/负载/内存/磁盘/服务/进程/端口) |
| Workbench CLI 会话 list/close |
保留的原插件语义
破坏性命令守卫(fail-closed):
rm -rf/shutdown/dd/systemctl stop等默认拒绝,须显式allow_destructive: true(会出现在 WorkBuddy 权限提示里)read_only 护栏:
read_only: true拒绝重定向/rm/docker 变更等 17 类写操作,并给出改写建议远端超时结算:
timed_out一律结算为exit_code=124(不显示成"成功")上传 sha256 校验:不一致即中止,坏包不落地
ANSI/进度帧清洗:去 ANSI/控制字符,并按
\r覆盖语义折叠上传/下载的进度帧(只保留最后一段真实结果),不再出现⠋ Preparing...残留;上传仅瞬时网络失败自动重试(指数退避)空会话归一:
ecs_session list在没有活动会话时返回{sessions: [], count: 0},而不是 CLI 原样的null
未移植(后续可加)
ecs_deploy / ecs_runbook / ecs_snapshot 编排引擎、伪会话(cd/export 继承)、detach 长任务、批量实例、可视化设置面板(dsh 专属 UI)。
开发与运行
cd /e/AiProject/workbench-ecs
node src/server.mjs # 前台直接起服务(stdio,通常由 WorkBuddy 拉起)
npm start # 同上
npm test # 冒烟测试:握手 + 8 个工具清单校验无需 npm install:运行时只用 Node 内置模块。package.json 中的 engines.node >= 18 是唯一的运行前提。
注册到 WorkBuddy
把 config/mcp.example.json 的片段合并进 ~/.workbuddy/mcp.json 的 mcpServers:
{
"workbench-ecs": {
"type": "stdio",
"command": "node",
"args": ["E:\\AiProject\\workbench-ecs\\src\\server.mjs"]
}
}然后在连接器管理页右上角"自定义连接器"里 Trust 该服务器。改动 src/server.mjs 后重启会话生效。
发布(维护者)
git tag v1.0.1 && git push origin v1.0.1打 tag 触发 .github/workflows/release.yml:测试与版本一致性校验 → npm publish --provenance → 确认版本真正上架 → 创建 GitHub Release。tag 必须与 package.json 的 version 一致,否则在第一步就被拦下。
npm 12 默认分阶段发布:npm publish 之后版本可能先进暂存队列,registry 上只出现 0.0.0-stage 占位(描述为 "Temporary package placeholder for staged publishing"),真实的 latest 要等放行才指向新版本。工作流会轮询最多 6 分钟;仍未公开则失败,并提示用 npm stage list + npm stage approve 人工批准(需 2FA)。
手动测试(协议层)
printf '%s\n' \
'{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"t","version":"0"}}}' \
'{"jsonrpc":"2.0","id":2,"method":"tools/list"}' \
| node src/server.mjs实测记录
2026-11-02(迁移前,旧路径):
ecs_exec对i-bp1xxxxxxxxxxxxxxxxx(示例生产机)执行hostname && uptime→ exit_code=0,118ms;破坏性守卫拒绝rm -rf;read_only 拒绝echo hi > /tmp/f。2026-11-14(迁移后,本路径):
npm test通过 ——workbench-ecs v1.0.0,协议2025-06-18,8 个工具全部就位。2026-11-14(改名后,经客户端 MCP 通道全量实测):
ecs_list按 status 过滤正常;ecs_find跨 22 地域检索(20 成功,ap-southeast-2 / ap-south-1 因本地 DNS 解析失败跳过并逐条报错);ecs_exec只读执行 83ms;ecs_diagnose分段体检正常;ecs_log读 400 字节并返回next_offset=400;ecs_upload→ecs_download往返 sha256 一致、内容无损;ecs_session list无活动会话时返回null。
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
This server cannot be deployed
Maintenance
Related MCP Connectors
Work on your own computers from Claude or ChatGPT: run commands, edit and search files.
Operate Linux, macOS and Windows from your LLM. Every action runs through an auditable allowlist.
Troubleshoot enrolled client PCs from your AI client, with human approval and an audit trail.
- DNS-DATA RemoteDeprecatedcom.dns-data
Connect AI assistants to authorized Windows and macOS devices for files, commands, and management.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables AI assistants to securely execute remote SSH commands, perform file transfers, and monitor system status through a standardized interface. It features robust security controls including command whitelisting, blacklisting, and credential isolation to prevent unauthorized operations.1023 npmMIT
- AlicenseNot gradedqualityBmaintenanceEnables AI assistants to perform operations on remote servers by reusing SSH sessions from a local desktop app, including command execution, file upload/download, and project document management with security controls.MIT
- AlicenseNot gradedqualityDmaintenanceEnables AI agents to manage remote servers via SSH, including command execution, multi-host batch operations, SFTP file transfer, background job handling, DevOps diagnostics, port tunneling, and safety guardrails like high-risk command blocking and read-only mode.MIT
- AlicenseAqualityBmaintenanceEnables AI clients to manage remote Linux workspaces over OpenSSH, including file read/write/edit, glob/grep search, shell command execution, and background job management, with authentication handled interactively in a terminal.9MIT