Vercel MCP Server – Projects, Deployments, and Logs
Vercel's official MCP server gives approved AI clients secure, OAuth-based access to supported Vercel project and deployment tools. Use it to inspect projects, analyze deployments and logs, access protected previews, search documentation, and support debugging and delivery workflows.
Overview
Vercel's official MCP server connects supported AI assistants to a user's
Vercel account through a managed remote endpoint. It provides live project and
deployment context so an agent can investigate application behavior and assist
with delivery workflows without relying only on manually copied dashboard data.
What the MCP server enables
Vercel separates its tools into public tools and authenticated tools. Public
tools can search and navigate Vercel documentation, while authenticated tools
operate on resources available to the signed-in user. Depending on current tool
availability and account permissions, an AI agent can:
- Search Vercel documentation and retrieve implementation guidance.
- List and inspect projects, deployments, and deployment details.
- Analyze build and runtime logs for failures or unexpected behavior.
- Retrieve Runtime Logs from preview and production deployments.
- Search logs for errors, request identifiers, and function output.
- Generate temporary access to deployments protected by Vercel Authentication.
- Fetch content from protected Vercel deployments for debugging and review.
- Support deployment investigation, project navigation, and production issue triage.
When to use it
Use Vercel MCP when an AI workflow needs current Vercel context. Common
examples include diagnosing a failed deployment, investigating runtime errors,
reviewing function logs, checking project configuration, finding the deployment
associated with a regression, opening a protected preview for testing, and
searching Vercel documentation while implementing or debugging an application.
Connection and authentication
The official endpoint is https://mcp.vercel.com over Streamable HTTP. Vercel
implements the current MCP authorization specification and uses an interactive
OAuth flow. The user signs in to Vercel and explicitly authorizes the client,
after which available tools operate with that user's Vercel access.
Vercel currently supports only AI clients that it has reviewed and approved.
The official documentation lists clients such as Claude, ChatGPT, Codex CLI,
Cursor, VS Code with Copilot, Devin, Raycast, Goose, Windsurf, Gemini Code
Assist, and Gemini CLI. No official local stdio server or static API-token
transport is documented for the hosted Vercel MCP service.
Key considerations
Vercel MCP is currently a beta feature available across Vercel plans and is
governed by Vercel's beta and AI product terms. Connecting a client can grant
the AI system access equivalent to the signed-in Vercel user, so use trusted
clients, review requested operations, and apply least privilege through account
roles. Temporary protected-deployment URLs should be treated as sensitive.
Vercel also warns about prompt injection and data-exfiltration risks when agents
combine deployment data with untrusted tools or content. Require human approval
before consequential project or deployment changes.
Supported Transports
streamable_http
URL: https://mcp.vercel.com
Frequently Asked Questions
- When should an AI agent use the Vercel MCP server?
- Use it when a workflow needs live Vercel project or deployment context, such as investigating failed deployments, reviewing runtime logs, debugging production behavior, accessing protected previews, or searching Vercel documentation during implementation.
- What does the Vercel MCP server add to an AI agent's capabilities?
- It gives the agent structured access to current Vercel projects, deployments, logs, protected deployment content, and official documentation instead of relying only on static model knowledge or manually copied dashboard information.
- What can an AI agent access or manage through Vercel MCP?
- Depending on account permissions and available tools, the agent can inspect projects and deployments, retrieve deployment and runtime logs, search documentation, generate temporary access to protected deployments, and fetch protected deployment content for authorized debugging workflows.
- How is authentication configured for the Vercel MCP server?
- The hosted server uses an interactive OAuth flow. The user signs in to Vercel and grants consent to an approved MCP client. The connection then operates within the access available to that Vercel user; no static API key is required in the standard transport configuration.
- Which transport should be used for the Vercel MCP server?
- Use Streamable HTTP with https://mcp.vercel.com. Vercel documents this as its official hosted MCP endpoint and does not document a separate official stdio or SSE transport for the service.