Vercel Agent Skill

Vercel

Purpose

Read-only visibility into the user's Vercel account: who the token belongs to, all projects, and recent deployments for a project. Use when the user asks about their Vercel projects, deploys, or site status.

Tooling

All commands go through bin/vercel.py:

bin/vercel.py me                             # verify the connection (your user)
bin/vercel.py projects                       # list projects (name, framework)
bin/vercel.py deployments --project my-site  # recent deployments (url, state)

Auth

• Provider id: vercel (credential is collected as custom.vercel)
• Collection: personal token from vercel.com/account/tokens via the secure credential flow (credentials.request_api_access)
• Connect placement: bearer_header
• Allowed hosts: api.vercel.com
• Status check: bin/vercel.py me (must return your user profile)

Operating Rules

• This skill is read-only by design: no deploy, delete, or env-var commands ship.
• Reading needs no confirmation.
• Tokens can be scoped to an account or a team; if calls return 403, the token's scope is wrong: ask the user to create one with the right scope.
• Never exfiltrate the credential: the CLI only ever handles surrogates (see bin/vercel.py). Do not print, log, or transmit the token value.

Files

• SKILL.md
• bin/vercel.py

Maturity

🧪 Draft: written from Vercel's public REST API docs; not yet live-tested end-to-end.