GitHub
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
作者提供的说明(英文原文)
MCP server for Dokku: let LLMs create, deploy, scale, configure and inspect Dokku apps.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
DOKKU_MCP_SSH_HOST, DOKKU_MCP_SSH_USER, DOKKU_MCP_SSH_PORT, DOKKU_MCP_SSH_KEY_PATH你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
链接指向作者自己的文件;我们不托管也不重新打包它。打开文件后,Claude Desktop 会进行安装。
在终端中运行这条命令
运行前请先填写空着的值。
浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
这个服务器在你的电脑上运行,而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
你的选择通过 Cookie 保存在这台设备上,无需账户。
DOKKU_MCP_SSH_USER envDOKKU_MCP_SSH_PORT envDOKKU_MCP_SSH_KEY_PATH env这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
它在你的电脑上运行;安装后,工具列表会显示在你的 AI 应用中。
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过 GitHub 账户 dokku-MCP 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
它在你的电脑上运行;我们只能读取互联网端点的工具列表
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 1 项适用的基础检查中通过了 1 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
你需要填写:DOKKU_MCP_SSH_HOST
它在你的电脑上运行。你的 AI 应用会在需要时启动它,数据不会经过作者的服务器(除非它自己调用在线服务)。
GitHub 仓库最后更新于–,最新版本(0.5.0)于29天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 MCPB。
同一分类中有:GitHub, Desktop Commander, Cloudflare MCP
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
MCP server for terminal commands, file operations, and process management
Cloudflare MCP servers
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Model Context Protocol (MCP) server for Dokku, written in Go.
Version: see releases (dokku-mcp --version)
This server exposes Dokku's management capabilities through the standardized Model Context Protocol (MCP), allowing Large Language Models (LLMs) to interact with and manage a Dokku instance.
⚠️ Early Development: This project is in its early stages. Breaking changes are expected, and it is not recommended for production use.
Follow Installation or grab a prebuilt binary to get started in minutes with cursor, claude-code, goose and all agentic tools which support mcp.
Or make start if you already have go locally.
For a quick tour of the server without wiring up a full MCP client, use the dedicated make target:
make inspect
It builds the binary (if needed), launches the MCP Inspector CLI, and connects it to the server in stdio mode. Inspector prints a local URL—open it in your browser to browse resources, prompts, and tools, or to issue ad-hoc calls. This is a great way to validate your setup before wiring Dokku MCP into Cursor, Claude, other IDE or internal tools.
The server can be used with any MCP-compatible client.
Add the following to your zed, .cursor/mcp.json, windsurf, claude_desktop_config.json or .mcp.json:
{
"mcpServers": { "dokku": { "command": "/home/user/dokku-mcp/build/dokku-mcp", "args": [], "env": { "DOKKU_MCP_SECURITY_BLACKLIST": "destroy,uninstall,remove", "DOKKU_MCP_SSH_HOST": "127.0.0.1", "DOKKU_MCP_SSH_KEY_PATH": "/home/user/.ssh/id_rsa", "DOKKU_MCP_SSH_PORT": "22", "DOKKU_MCP_SSH_USER": "dokku" } } } }
*Remember to replace the `command` path with the absolute path to the binary.*
Tools are grouped by plugin. Plugins tied to a Dokku plugin (for example Let's Encrypt) only appear when that plugin is installed. Every tool carries MCP annotations (title, read-only, destructive, idempotent), and newer tools return structured content with an output schema.
| Area | Tools |
|---|---|
| Applications | create_app, deploy_app, rollback_app, scale_app, configure_app, get_app_status, restart_app, stop_app, start_app |
| Deployments | get_deployment_status, list_deployments |
| Logs | get_runtime_logs, follow_runtime_logs (live, streamed as progress notifications), get_failed_deploy_logs |
| Datastores | list_services, create_service, get_service_info, link_service, unlink_service, get_service_logs, destroy_service (postgres, mysql, mariadb, redis, mongo, rabbitmq, memcached, clickhouse, elasticsearch) |
| Domains | get_app_domains, add_app_domain, remove_app_domain, list_global_domains, add_global_domain |
| HTTPS | get_letsencrypt_status, enable_letsencrypt, disable_letsencrypt, set_letsencrypt_email |
| Access | list_ssh_keys, add_ssh_key, remove_ssh_key, get_registry_report, registry_login, registry_logout |
| Server | get_server_logs (opt-in with expose_server_logs) |
Resources include dokku://onboarding/quickstart (start here), dokku://onboarding/capabilities, dokku://onboarding/intent-map, dokku://core/server/info and dokku://app/{name}/logs. The app_doctor prompt walks through diagnosing an application.
Deployments are asynchronous: deploy_app returns a deployment_id as soon as the code is synced, and get_deployment_status reports the build outcome and the tail of the build log.
The server runs Dokku commands over SSH with the configured key, so treat it like that key.
security.allowlist restricts the server to matching commands ("apps:", "config:*", "logs", ...); security.blacklist blocks commands containing a pattern (destroy, uninstall, ...). Both can be set with DOKKU_MCP_SECURITY_ALLOWLIST / DOKKU_MCP_SECURITY_BLACKLIST as comma-separated lists.destroy_service requires the service name to be repeated; add_ssh_key refuses names containing admin (which Dokku lets add further keys) unless allow_admin is set.Submit an issue for re-priorizing proposal
apps:list, apps:info, apps:create, apps:exists, apps:report, config:show, config:set --encoded, ps:scale, ps:report, ps:rebuild, ps:restart, ps:stop, ps:start, git:sync, logs (including -t follow), logs:failed, domains:report, domains:add, domains:remove, domains:add-global, letsencrypt:*, :create/info/list/link/unlink/links/logs/destroy, ssh-keys:list/add/remove, registry:login/logout/report, plugin:list, version, proxy:report, scheduler:report, git:report.ports:*, checks:*, storage:*, cron:*, datastore backups.Open an issue or read Contributing.
Need help with the dokku-mcp project? Goose AI is available to assist you! Simply tag your GitHub issues or pull requests with the :goose label, and Goose AI will automatically be notified to help with development, debugging, feature implementation, and other project-related tasks.
Example of Goose AI in action Another example
Download the latest release for your platform:
# Linux (amd64; also available: arm64, arm)
curl -L -o dokku-mcp https://github.com/dokku-mcp/dokku-mcp/releases/latest/download/dokku-mcp-linux-amd64 chmod +x dokku-mcp sudo mv dokku-mcp /usr/local/bin/
curl -L -o dokku-mcp https://github.com/dokku-mcp/dokku-mcp/releases/latest/download/dokku-mcp-darwin-amd64 chmod +x dokku-mcp sudo mv dokku-mcp /usr/local/bin/
Compressed archives (dokku-mcp- - .tar.gz) are also published with each release.
Each stable release also ships dokku-mcp.mcpb, an MCP Bundle for macOS and Linux that MCP clients such as Claude Desktop install in one step, asking for the Dokku host, SSH user, port and key. The same bundle is published to the official MCP Registry as io.github.dokku-MCP/dokku-mcp.
dokku-mcp --version
If you prefer to build from source:
git clone https://github.com/dokku-mcp/dokku-mcp.git
cd dokku-mcp make build
The server can be configured in two ways: using a config.yaml file or via environment variables.
Create a configuration file at one of the following locations:
/etc/dokku-mcp/config.yaml~/.dokku-mcp/config.yamlconfig.yaml in the same directory as the binary.Here is a minimal config.yaml example:
ssh:
host: "your-dokku-host.com" user: "dokku"
log_level: "info"
For a full list of available options, please refer to the config.yaml.example file.
All configuration settings can be overridden with environment variables prefixed with DOKKU_MCP_. Nested keys use underscores (ssh.host → DOKKU_MCP_SSH_HOST) and lists are comma-separated. For example:
export DOKKU_MCP_SSH_HOST="your-dokku-host.com"
export DOKKU_MCP_SSH_USER="dokku" export DOKKU_MCP_LOG_LEVEL="debug" export DOKKU_MCP_SECURITY_ALLOWLIST="apps:,config:,ps:,git:sync,logs,version"
Once configured, you can run the server:
dokku-mcp
The server will start and be ready to accept connections from an MCP client.
For development and testing without needing a remote Dokku instance, you can run a local Dokku server using Docker.
Prerequisites:
This command will download the necessary Docker images and configure the local Dokku instance. It only needs to be run once.
make setup-dokku
make dokku-start
make dokku-stop
When the local Dokku container is running, the MCP server (with default config) should be able to connect to it. You can run integration tests against this local instance.
The server supports two transport modes for clients:
DOKKU_MCP_TRANSPORT_TYPE=sse dokku-mcp
For SSE transport, CORS is handled by default with Access-Control-Allow-Origin: *. For remote deployments, you can enable custom CORS middleware:
transport:
type: sse cors: enabled: true allowed_origins:
See docs/CORS.md for detailed configuration options and security best practices.
This section is for developers who want to contribute to the project or modify the source code.
git clone https://github.com/dokku-mcp/dokku-mcp.git
cd dokku-mcp
This command inst
只显示了开头部分;其余内容请在仓库中阅读