Vercel MCP now renders deployments as an interactive card

Vercel's MCP server, which lets AI assistants look up and act on a Vercel account, can now show a deployment as an interactive card inside OpenAI's ChatGPT and other chat clients that support MCP Apps (Vercel changelog). MCP, the Model Context Protocol, is the standard way chat clients connect to outside tools, and a tool normally returns text. MCP Apps is an extension that lets a tool point to an HTML interface, which the chat client renders in an isolated frame; the interface cannot reach the server directly and sends its requests through the client, which decides what to forward (Vercel guide to MCP Apps).

The card shows a deployment's live status, build logs, and source and commit details, with a Visit button once the deployment's URL is ready. From the card, a user can send the logs into the chat so the model can help debug, open a fullscreen view of the logs, or switch between projects when a repository is linked to several. The open_deployments tool behind the card finds deployments by project, repository or commit and can filter by environment. Chat clients without MCP Apps support receive the deployment state, preview URL and commit as text instead.

This is Vercel MCP's first MCP App. It follows a July change that let the same server deploy code from a chat and return a URL while the build finished in the background; together, the two let a user deploy and then follow the build in the same conversation, in clients that support MCP Apps. ChatGPT connects to Vercel MCP through its Connectors setup.