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
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
E2B
Using MCP to run code via e2b.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.