Shipyard

Shipyard is an MCP server that integrates the Shipyard CLI with Model Context Protocol clients to inspect and manage ephemeral cloud environments. Built for DevOps engineers, software developers, and QA teams using the Shipyard platform, it exposes native capabilities directly to AI assistants such as Claude. The server connects your AI client to Shipyard's infrastructure APIs, enabling users to monitor and orchestrate dynamic preview environments through natural conversation. Using this server, assistants can list and filter active or deleted environments by branch, repository, or pull request. It allows operators to control lifecycle actions including starting, restarting, cancelling, rebuilding, and reviving environments. Assistants can also inspect container services, stream service logs, manage storage volumes, create or restore volume snapshots, and switch across multiple organization contexts without manually running terminal commands.

Category: Cloud & Infrastructure

Tags: ci-cd, environments, infrastructure, shipyard, testing

Visit Shipyard

How to install and configure Shipyard

1. Install the Shipyard CLI on your system using your preferred package manager (e.g., via Homebrew with brew tap shipyard/tap && brew install shipyard or the official install script curl https://www.shipyard.sh/install.sh | bash). 2. Authenticate the CLI by running shipyard login or setting the SHIPYARD_API_TOKEN environment variable from your profile page. 3. Add the MCP server to Claude by running: claude mcp add shipyard --env SHIPYARD_API_TOKEN=your-token-here --env SHIPYARD_ORG=your-org-name -- shipyard mcp serve (or omit the environment flags if credentials are already configured via the CLI). 4. Alternatively, for Codex CLI, edit ~/.codex/config.toml to register command = "shipyard" with args = ["mcp", "serve"] and your corresponding environment variables.

What you can do with Shipyard

• Filter and locate active or deleted ephemeral environments by pull request number, Git branch, or repository name. - Control deployment lifecycles by stopping, restarting, rebuilding with the latest commit, or reviving deleted preview environments. - Inspect running containerized services and retrieve operational logs directly from specific application containers for troubleshooting. - Manage persistence by creating volume snapshots, restoring data from specific sequence numbers, or resetting volumes to their baseline. - Switch the active Shipyard organization context and retrieve workspace configurations during multi-tenant infrastructure operations.

Key facts

• https://github.com/shipyard/shipyard-cli
• Cloud & Infrastructure, DevOps, CI/CD & Version Control
• ci-cd, environments, infrastructure, shipyard, testing

Part of MCP Servers

What is the Shipyard MCP server?

The Shipyard MCP server is a built-in feature of the Shipyard CLI that exposes ephemeral environment management tools to AI assistants via the Model Context Protocol. It allows LLMs to query environments, read service logs, create volume snapshots, and trigger environment lifecycle commands.

How do I install the Shipyard MCP server?

You install it by installing the Shipyard CLI on Linux, macOS, or Windows. On macOS, you can use Homebrew via brew tap shipyard/tap and brew install shipyard. Once installed, start the server inside your MCP client by calling shipyard mcp serve.

Which MCP clients work with Shipyard?

The Shipyard CLI documentation explicitly provides integration instructions for Claude using the claude mcp add command and for Codex CLI via its config.toml file. Any client supporting local command-based Model Context Protocol servers can also connect.

Can the Shipyard MCP server execute commands inside containers?

Tools like exec_service, port_forward, and telepresence_connect are limited inside the MCP implementation. Rather than running interactive terminal sessions directly through the AI model, these tools return instructional help directing users to execute the native CLI commands instead.

What can Shipyard MCP do with volumes?

The server provides five dedicated volume tools. It allows AI assistants to list storage volumes attached to an environment, reset a volume to its clean state, list existing snapshots, create new snapshots, and restore an environment from a specific snapshot sequence.