Vercel MCP Server

What it does

The Vercel MCP Server connects AI coding agents to your Vercel account. Agents can list teams and projects, get project details, list deployments, analyze deployment build logs, and search Vercel documentation. Authentication uses OAuth at https://mcp.vercel.com, so no API keys are stored in config files.

Available tools

Tool	What it does
list_teams	List Vercel teams you have access to
list_projects	List projects in a team
get_project	Get project configuration and metadata
list_deployments	List deployments for a project
get_deployment	Get deployment details
get_build_logs	Retrieve build logs for a deployment
search_docs	Search Vercel documentation

Setup by tool

Cursor

Install using the add-mcp command:

npx add-mcp https://mcp.vercel.com

This detects installed AI clients and configures them. Or add manually to .cursor/mcp.json:

{
  "mcpServers": {
    "vercel": {
      "url": "https://mcp.vercel.com"
    }
  }
}

On first use, you will be prompted to log in to Vercel via OAuth.

Claude Code

Add to your Claude configuration file:

{
  "mcpServers": {
    "vercel": {
      "url": "https://mcp.vercel.com"
    }
  }
}

VS Code / GitHub Copilot

Add the same url to your MCP client. OAuth will complete in the browser when you first invoke the server.

When to use this

• Deployment debugging: Pull build logs into the IDE when a deployment fails
• Project context: Agents understand project structure and environment before making changes
• Docs lookup: Search Vercel documentation without leaving the coding flow
• Deployment status: Check deployment state after pushing changes or merging PRs

Security considerations

• OAuth grants access to your Vercel account; use the same care as with the Vercel dashboard
• The server respects your Vercel permissions; it cannot access projects you cannot access
• Do not share MCP config that includes OAuth tokens; tokens are stored by the client
• Review tool calls before execution when your client supports manual approval