captain-tool
v0.0.26
Published
MCP server connecting Claude Desktop and VS Code Copilot to Captain Cloud
Readme
captain-tool
An MCP server that connects your AI coding agent to Captain — a release train management and SDLC orchestration platform.
Once installed, your AI agent can interact with your Captain workspace directly in conversation: checking pipeline status, reviewing deployments, managing artifacts, and more.
Supported clients
- Claude Desktop
- Claude Code (CLI)
- VS Code + Copilot
- Cursor
- Windsurf
Requirements
- Node.js 16+
- A Captain account
- One of the supported AI clients above
Installation
npm install -g captain-toolSetup (recommended)
Run the interactive setup wizard to automatically configure your AI client(s):
captain-tool setupThe wizard will detect your platform, find the binary, and write the correct MCP config for whichever client(s) you select. Supports all 5 clients above in one step.
Manual setup
Claude Desktop
Edit %APPDATA%\Claude\claude_desktop_config.json (Windows) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):
macOS / Linux:
{
"mcpServers": {
"captain-tool": {
"command": "captain-tool",
"env": {
"CAPTAIN_API_URL": "https://app.getcaptain.dev/"
}
}
}
}Windows (requires cmd /c wrapper so the shell can resolve the .cmd shim):
{
"mcpServers": {
"captain-tool": {
"command": "cmd",
"args": ["/c", "captain-tool"],
"env": {
"CAPTAIN_API_URL": "https://app.getcaptain.dev/"
}
}
}
}Restart Claude Desktop to activate.
Claude Code (CLI)
claude mcp add captain-tool -- cmd /c captain-toolVS Code + Copilot
Or edit %APPDATA%\Code\User\mcp.json (Windows) / ~/.config/Code/User/mcp.json (Linux) / ~/Library/Application Support/Code/User/mcp.json (macOS):
{
"servers": {
"captain-tool": {
"type": "stdio",
"command": "captain-tool",
"env": {
"CAPTAIN_API_URL": "https://app.getcaptain.dev/"
}
}
}
}Reload VS Code (Ctrl+Shift+P → "Developer: Reload Window") to activate.
Cursor
Edit ~/.cursor/mcp.json:
{
"mcpServers": {
"captain-tool": {
"command": "captain-tool",
"env": {
"CAPTAIN_API_URL": "https://app.getcaptain.dev/"
}
}
}
}Restart Cursor to activate.
Windsurf
Edit ~/.codeium/windsurf/mcp_config.json (macOS/Linux) or %USERPROFILE%\.codeium\windsurf\mcp_config.json (Windows):
{
"mcpServers": {
"captain-tool": {
"command": "captain-tool",
"env": {
"CAPTAIN_API_URL": "https://app.getcaptain.dev/"
}
}
}
}Restart Windsurf to activate.
First use
Once connected, ask your AI agent:
"Check captain status and get me set up."
It will walk you through logging in and selecting your workspace. After that, you can ask things like:
Show me all my pipelines.
What's the current state of my active workspace?
List any unhealthy connectors.
What components do we have?
Show me the deployment history for component comp-xyz.Configuration
| Variable | Default | Description |
|----------|---------|-------------|
| CAPTAIN_API_URL | https://app.getcaptain.dev/ | Captain backend URL |
| CAPTAIN_DB_PATH | platform default | Local state and auth token |
| CAPTAIN_REQUEST_TIMEOUT_SECONDS | 30 | HTTP timeout per request |
Learn more
Visit app.getcaptain.dev to learn more about Captain.
