MCP Server

Jetpacked exposes a remote MCP (Model Context Protocol) server at mcp.jetpacked.ai. This lets you deploy, monitor, and manage your projects directly from any AI tool that supports MCP — Claude Desktop, Cursor, Windsurf, VS Code Copilot, and more.

You can ask your AI assistant things like "deploy my repo", "check the deployment logs", or "create a new project from this GitHub repo" and it will do it through Jetpacked.

What you can do

The MCP server exposes the following tools to your AI assistant:

Tool	What it does
list_projects	List all your projects with their status and URLs
get_project	Get full details of a specific project
list_github_repos	List GitHub repositories connected to your account
list_branches	List branches of a specific repo
create_github_repo	Create a new GitHub repo for the user
analyze_project	Clone a repo, run the detection engine, return a draft with detected config
deploy_project	Create a new project and trigger the first deployment
redeploy_project	Redeploy an existing project (re-analyzes the repo)
get_deployment	Get the status and stages of a deployment
get_deployment_logs	Fetch full build logs for a deployment

Project creation is a two-step process: analyze_project runs the detection engine and returns a draft token, then deploy_project finalises the configuration and triggers the first deployment. This gives your AI assistant the chance to present the detected configuration to you before committing.

Step 1 — Get an API key

Go to Settings → API Keys and create a key. Give it a name that identifies which tool you're using it with (e.g. "Claude Desktop"). Copy the key — it is only shown once.

Step 2 — Connect your AI tool

Pick your tool and copy the config or setup details. For Claude Desktop and Cursor you can use the install button to open the app directly — you'll still need to paste your API key in afterwards. For OpenAI Codex, add the Jetpacked MCP server to your local Codex config.

{
  "mcpServers": {
    "jetpacked": {
      "url": "https://mcp.jetpacked.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Deploying a new project via MCP

Here is a typical conversation flow with your AI assistant:

1. Ask it to "deploy my GitHub repo username/my-app on the main branch"
2. It will call analyze_project — this clones the repo and runs the detection engine
3. It will present the detected framework, services, and any required environment variables
4. Confirm, provide any missing env vars, then ask it to proceed
5. It calls deploy_project with the draft token and starts the deployment
6. You can ask it to "check the deployment logs" to monitor progress

If you don't have a GitHub repo yet, ask the assistant to "create a GitHub repo called my-app" first — it will use create_github_repo and then walk you through the deployment.

Security

API keys grant full account access. Each key is stored as a one-way SHA-256 hash — the plaintext is never persisted. Revoke keys you no longer use in Settings → API Keys.