Vercel

Search Vercel docs, inspect deployments, read logs, deploy to production and buy domains.

Add to Agentman Studio

Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.

Paste it into any MCP client. Setup docs

Resources

What the Vercel connector does

The Vercel MCP server connects the Vercel frontend cloud to Claude, ChatGPT and any approved MCP-compatible agent. It searches Vercel's documentation, lists projects and deployments, reads build and runtime logs, deploys a file tree straight to production, and registers domains against your team's card. Sign-in is OAuth through Vercel's own login.

Verified connector

Listed by Anthropic as a partner connector in its Connectors Directory.

Connection checked by Agentman on .

Anthropic states this reflects the level of review a connector received, not a security audit.

Vercel tools (7)

  • search_vercel_documentation
  • list_projects
  • get_project
  • list_deployments
  • get_deployment
  • get_deployment_events
  • list_teams

Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.

Limits

  • We could not verify the tool list live. https://mcp.vercel.com/ returns 401 to every unauthenticated request, so the 32-tool figure is Vercel's documentation and the 7-tool figure is Anthropic's snapshot. Neither is our own handshake.
  • The two sources disagree by 25 tools, and one listed name — get_deployment_events — has no documented counterpart. Treat the directory listing as an undercount.
  • Vercel's overview page describes public tools available without authentication. Our probe found no such path: an unauthenticated GET and an unauthenticated initialize both returned 401 with a Bearer challenge on 2026-08-19. Documentation search may be public once a session exists, but the endpoint itself is not.
  • No capability scopes exist. The consent screen cannot offer read-only. Restriction comes from the Vercel role you sign in with.
  • No documented org-level control over the connector. We found no Vercel setting that lets an admin block or restrict Vercel MCP for a team.
  • Deploying has no server-side confirmation. The gate is your MCP client's, not Vercel's.
  • Purchases are non-refundable and charge the team's payment method immediately, by Vercel's own statement.
  • Build-log redaction has a 32-character floor and applies only to variables explicitly marked sensitive.
  • Purchase tools are still rolling out and may be absent from your connection.
  • get_web_analytics aggregate queries are bounded by your plan's reporting window, and require Web Analytics to be enabled on the project.
  • get_runtime_errors looks back at most seven days.

Frequently asked questions

Can the Vercel MCP server deploy to production?

Yes. Vercel documents a deploy_to_vercel tool that takes a file tree and a target parameter, and Vercel's own description names production as an accepted target value. It creates the project if it does not exist, detects the framework and starts the build. Nothing server-side gates that call, so the only barrier is your MCP client's human-confirmation prompt, which Vercel tells you to enable.

Can Vercel MCP read my environment variable values?

No. None of the 32 tools Vercel documents reads, writes or lists environment variables at all, so a secret stored in a Vercel env var is not reachable through this connector. Separately, Vercel marks variables created with the Sensitive option as non-readable once created, which protects them even from the dashboard and REST API.

Do Vercel MCP tools leak secrets in build logs?

Partly protected, and the protection is conditional. Vercel replaces a sensitive environment variable value with REDACTED in build logs only when the value is 32 characters or longer, and always redacts VERCEL_AUTOMATION_BYPASS_SECRET and VERCEL_OIDC_TOKEN regardless of length. A short secret, or one not marked sensitive, is not redacted and get_deployment_build_logs will return it.

Can the Vercel MCP server spend money?

Yes. Vercel documents buy_pro, buy_credits, buy_addon and buy_domain, and states charges go to the team payment method immediately and are non-refundable. Every purchase needs a quote from get_purchase_quote first, and the buy call must echo back a signed idempotencyKey that expires in five minutes and must match the quoted terms exactly.

Why does Anthropic list only 7 Vercel tools?

The directory snapshot is stale. Anthropic's listing records 7 tools dated 2026-08-16, while Vercel's own tools reference dated 2026-07-23 documents 32. One listed name, get_deployment_events, does not appear in Vercel's documentation at all, and the closest current equivalent is get_deployment_build_logs.

Which Vercel teams can the connector reach?

All of them. Vercel states that connecting to Vercel MCP grants the AI system the same access as your Vercel user account, and list_teams enumerates every team you belong to. Almost every other tool takes a teamId parameter, so the model picks the team per call. Restrict reach by signing in with a narrower Vercel role, not through the consent screen.

Does the Vercel MCP connector work with any MCP client?

No. Vercel states it only supports AI clients that it has reviewed and approved, and publishes the current list: Claude Code, Claude.ai and Claude Desktop, ChatGPT, Codex CLI, Cursor, VS Code with Copilot, Devin, Raycast, Goose, Windsurf, Gemini Code Assist and Gemini CLI. An unlisted client is not simply unconfigured, it is not permitted to connect.

Sources

  • Vercel MCP overview — https://vercel.com/docs/agent-resources/vercel-mcp (fetched 2026-08-19; the directory's published URL https://vercel.com/docs/mcp/vercel-mcp redirects here) · retrieved 2026-08-19
  • Vercel MCP tools reference — https://vercel.com/docs/agent-resources/vercel-mcp/tools (fetched 2026-08-19; page states last updated 2026-07-23) · retrieved 2026-08-19
  • Vercel sensitive environment variables — https://vercel.com/docs/environment-variables/sensitive-environment-variables (fetched 2026-08-19) · retrieved 2026-08-19
  • Vercel access roles — https://vercel.com/docs/rbac/access-roles (fetched 2026-08-19) · retrieved 2026-08-19
  • Live 401 probe of https://mcp.vercel.com/ — 2026-08-19 · retrieved 2026-08-19
  • RFC 9728 protected-resource descriptor — https://mcp.vercel.com/.well-known/oauth-protected-resource (fetched 2026-08-19) · retrieved 2026-08-19
  • RFC 8414 authorization-server metadata — https://vercel.com/.well-known/oauth-authorization-server (fetched 2026-08-19) · retrieved 2026-08-19
  • Anthropic connector directory — https://claude.ai/directory/7eb42afe-0087-4493-a105-da2b021d5c03 (snapshot 2026-08-16) · retrieved 2026-08-16
  • https://vercel.com/robots.txt (fetched 2026-08-19) carries Content-Signal: search=yes, ai-input=yes, ai-train=no — an explicit publisher opt-in to AI synthesis, honoured here. · retrieved 2026-08-19

More Developer Tools connectors

  • Base44

    Build and edit Base44 apps from an AI assistant, and query data your app's users submit.

  • CircleCI

    Diagnose failed builds, read job logs, inspect test results and rerun or roll back deployments from your AI assistant.

  • Claude Managed Agents

    Create, run, monitor and debug Anthropic's Claude Managed Agents from inside Claude itself.

Browse all MCP connectorsClaude and ChatGPT setup guideBrowse AI skills

Product names and logos belong to their owners. Agentman is not affiliated with a connector's maker unless the page says the connector is built and run by Agentman.

Use it in an agent

Put Vercel to work.

Connect Vercel once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.

Open in Agentman Studio
Call (650) 285-1019Our AI receptionist answers.