Source: vercel-labs/agent-skills · Maintainer: Vercel · License: MIT (per the repo README; no LICENSE file in the repo) · Checked: 24 October 2026
What is the Vercel CLI skill?
The Vercel CLI skill, vercel-cli-with-tokens, teaches your coding agent to deploy and manage Vercel projects through the Vercel CLI using an access token instead of an interactive vercel login. It is for setups where nobody can click through a browser login: CI jobs, remote dev boxes, cloud agents and containers.
It lives in the vercel-cli-with-tokens folder of vercel-labs/agent-skills, Vercel's official skill collection. The description's first sentence reads: "Deploy and manage projects on Vercel using token-based authentication." It then names triggers such as "deploy to vercel", "set up vercel" and "add environment variables to vercel".
As of 24 October 2026, its skills.sh page listed about 134,400 installs (snapshot, 24 October 2026). The folder holds one file, SKILL.md, last updated in April 2026.
What does it do?
The skill is model-invoked, so the agent can load it when you ask to deploy, set up Vercel or add environment variables while a token is involved. It works in a fixed order.
1. Find the token. It checks, in order: VERCEL_TOKEN already in the environment, VERCEL_TOKEN= in a .env file, any Vercel-looking variable in .env (which it exports as VERCEL_TOKEN), and finally asks you. Once exported, the agent must not pass it as --token. The Vercel CLI global options document both methods and recommend the environment variable because it "avoids exposing the token in command-line arguments, which can be visible in process lists and logs."
2. Find the project and team. It looks for VERCEL_PROJECT_ID and VERCEL_ORG_ID, or pulls the team slug out of a project URL. With both IDs set, the CLI can skip vercel link. The skill notes the two must be set together.
3. Deploy. Preview by default:
| What you have | What the agent runs |
|---|---|
| Token and project ID | vercel deploy -y --no-wait, add --scope <team-slug> if needed |
| Token and team, no project ID, git remote | vercel link --repo --scope <team-slug> -y, then asks before a git push |
| Token and team, no git remote | vercel link --scope <team-slug> -y, then vercel deploy |
| A repo URL only | git clone, link, then push or CLI deploy |
Production uses --prod, only when you ask for it.
4. Manage the project. The skill includes ready commands for vercel env add, env ls, env pull and env rm; vercel ls --format json; vercel inspect with --logs for build logs; vercel logs for runtime logs; and vercel domains ls and domains add.
5. Stripe Projects plan changes. A short section covers upgrading or downgrading a Vercel plan managed by Stripe Projects, with a rule to ask the user first because "upgrades bill a real card, downgrades remove seats."
It ends with a working agreement: check the environment before asking for a token, default to preview, ask before pushing, never edit .vercel/ files by hand, don't curl deployed URLs, use -y to avoid interactive prompts, and use --format json when structured output helps.
How do you install it?
The skills.sh page gives this command. The skills CLI matches --skill against the frontmatter name, which here matches the folder:
npx skills add https://github.com/vercel-labs/agent-skills --skill vercel-cli-with-tokens
Add -g for all projects or -a claude-code to target Claude Code. A global install lands in ~/.claude/skills/vercel-cli-with-tokens/, one of the standard Claude Code skill locations. Codex and Cursor read the same SKILL.md format; see Codex skills and Cursor skills.
The repo also stores a vercel-cli-with-tokens.zip with the same single SKILL.md. The README does not describe a claude.ai route for this skill, and it needs a shell that can run the Vercel CLI.
Then give the agent a token. Vercel's token docs explain how to create one at vercel.com/account/tokens with Full Account, Team or Project scope.
How do you use it?
Set VERCEL_TOKEN (and ideally VERCEL_ORG_ID and VERCEL_PROJECT_ID) in the environment or .env, then ask. Example prompts (suggestions, not quotes from the repo):
Deploy this to Vercel as a preview. The token is in .env.Add STRIPE_SECRET_KEY to the production environment on Vercel.The last Vercel build failed. Show me the build logs and fix the cause.Add docs.example.com to this Vercel project.Clone github.com/acme/site, link it to the acme team on Vercel and deploy a preview.
Requirements: the Vercel CLI (npm install -g vercel), a Vercel access token, and Node.js for the CLI. The skill says vercel inspect --logs needs Vercel CLI v35 or later.
How is it different from the Vercel plugin's vercel-cli skill?
Searching for a "Vercel CLI skill" can also turn up the vercel-cli skill in Vercel's separate vercel/vercel-plugin package. They overlap but aim at different jobs.
vercel-cli-with-tokens |
vercel-cli in vercel-plugin |
|
|---|---|---|
| Where | vercel-labs/agent-skills | vercel/vercel-plugin |
| Install | npx skills add ... --skill vercel-cli-with-tokens |
npx plugins add vercel/vercel-plugin (whole plugin) |
| Focus | Token auth without vercel login |
General CLI guidance: deploys, env vars, logs, metrics, debugging, domains, flags |
| Files | One SKILL.md | Part of a plugin with many skills, agents, commands and hooks |
Pick vercel-cli-with-tokens when the agent runs somewhere without a browser login. Pick the plugin when you want broad Vercel knowledge across products in a supported agent.
Tips and limits
From the SKILL.md and Vercel's docs, not from testing:
- The token prefix hint may not match your token. The skill says Vercel tokens "typically start with
vca_". Vercel's token docs say "Personal access tokens begin with the prefixvcp_". The skill's.envsearch looks for any line containing "vercel", so a token stored under another name is still found. - Scope the token narrowly. Vercel's docs advise: "Choose the narrowest scope that supports your task." A project-scoped token limits what the agent can touch.
- Stripe plan commands can spend money. On 24 October 2026 the skills.sh page showed Gen Agent Trust Hub: Pass, Socket: Pass and Snyk: Warn. Snyk's audit dated 15 September 2026 lists one medium issue, W009 "Direct money access capability detected", because of the Stripe Projects upgrade and downgrade commands. The skill tells the agent to ask before any paid or destructive plan change.
- Plan details in the skill can go stale. The Stripe section lists Pro plan figures; check Vercel's Pro plan page for current terms.
- No scripts. The skill runs only CLI and git commands the agent types, and it asks before any git push.
The MIT License lets you copy and adapt the skill as long as the notice stays with it.
Save it in your Kipt library
In Kipt, open Skills and paste the skill's GitHub folder link into the import box. Kipt reads the SKILL.md; for a repo with many skills, like this one, it shows a list to pick from. You can also drop the folder or a .zip. Your skills live in your own private GitHub repo and install into Claude Code or Codex with one command:
npx skills add https://github.com/<you>/<repo>.git --all -g
Kipt is not affiliated with Vercel.
Similar skills
- Deploy to Vercel skill: the interactive version, with a no-login fallback for claude.ai and Codex.
- Vercel optimize skill: audits a deployed project's cost and performance from Vercel metrics.
- Claude Code GitHub Actions: run Claude Code in CI, where token-based Vercel access fits naturally.
- Find skills skill: search skills.sh for other deployment and platform skills.