bcdocker
MCP server for managing Business Central Docker containers, enabling AI assistants to list, create, test, and manage BC sandboxes via natural language.
README
bcdocker
MCP server and CLI for Business Central Docker container management. Run via npx -- no install needed.
npx bcdocker list # list containers
npx bcdocker create # create a BC sandbox
npx bcdocker test bcsandbox # run AL tests
Quick start
As MCP server (AI assistants)
Zero-install -- add to your IDE config and your AI can manage BC containers directly:
Cursor (.cursor/mcp.json):
{
"mcpServers": {
"bcd": {
"command": "npx",
"args": ["-y", "bcdocker", "mcp"]
}
}
}
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"bcd": {
"command": "npx",
"args": ["-y", "bcdocker", "mcp"]
}
}
}
VS Code / Copilot (.vscode/mcp.json):
{
"servers": {
"bcd": {
"type": "stdio",
"command": "npx",
"args": ["-y", "bcdocker", "mcp"]
}
}
}
Once configured, ask your AI assistant things like:
- "List my BC containers"
- "Create a new BC sandbox with version 26.0"
- "Show me the apps in bcsandbox"
- "Run tests in my container"
- "Compile and publish my AL project"
As CLI
Via npx (no install):
npx bcdocker list
npx bcdocker info bcsandbox
npx bcdocker create --name bcsandbox --type sandbox --bc-version 26.0 --country w1 --bypass-cdn
npx bcdocker start bcsandbox
npx bcdocker stop bcsandbox
npx bcdocker apps bcsandbox --publisher Microsoft
npx bcdocker test bcsandbox --codeunit 50100
Global install (puts bcd on your PATH):
npm i -g bcdocker
bcd list
bcd create --name mybc --bc-version 26.0
bcd test mybc
Prerequisites
- Node.js 18+
- Windows PowerShell 5.1 (BC management cmdlets require it)
- Docker Desktop in Windows containers mode
The npm package bundles the PowerShell module in ps/. Override with BCD_MODULE_PATH if needed.
CLI commands
| Command | Description |
|---|---|
list / ls |
List all BC containers with status |
info <container> |
Show version, status, and endpoints |
create |
Create a new BC container (5-30 min) |
remove <container> |
Remove a container |
start <container> |
Start a stopped container |
stop <container> |
Stop a running container |
restart <container> |
Restart a container |
open <container> |
Open BC Web Client in browser |
apps <container> |
List apps (optional --publisher) |
install <container> <appFile> |
Install a .app file |
uninstall <container> <app> <publisher> |
Uninstall an app |
publish <container> <folder> |
Compile and publish an AL project |
test [container] |
Run AL tests (optional --codeunit, --function, --app) |
toolkit <container> |
Import test toolkit (--full for all) |
license <container> <file> |
Import a license file |
MCP tools
The same 15 operations exposed as MCP tools for AI assistants:
list-containers container-info create-container remove-container start-container stop-container restart-container open-webclient list-apps install-app uninstall-app publish-project run-tests import-test-toolkit import-license
Naming
| Term | What it is |
|---|---|
bcdocker |
This npm package and GitHub repo. |
bcd |
The CLI command after global install (npm i -g bcdocker). |
bcdocker-toolkit |
The PowerShell + Windows UI repo (module BCDocker). Different repo. |
Related
- bcdocker-toolkit -- PowerShell module and Windows UI for Business Central Docker containers
Migrating from 1.x to 2.0
Breaking changes in 2.0:
- MCP server invocation: add
"mcp"arg to your MCP client config. The previous config"args": ["-y", "bcdocker"]ran the CLI (which showed help instead of starting the MCP server). New:"args": ["-y", "bcdocker", "mcp"]. bcd createflags: the--versionflag (which actually meant "artifact type") is replaced by--type sandbox|onpremand-v, --bc-version <version>. Previouslybcd create --version sandbox→ nowbcd create --type sandbox. A specific BC version like28.0now works:bcd create --bc-version 28.0.- MCP
create-containerschema: theversionfield is split intotype(enum: sandbox, onprem) andbcVersion(string, e.g."28.0"or empty for latest). - MCP tool responses: handlers now surface stderr and non-zero exit codes as
Failed:\n...text instead of returning the confident fallback message. AI clients will get actionable failure details instead of spurious "complete" messages.
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。