Vercel Plugin Skills - Setup Guide¶
Source: vercel/vercel-plugin (295⭐, NOASSERTION)
Skill family: vercel/vercel-plugin (11 indexed skills; 43 in-repo under skills/)
Combined Installs: ~4,768 across indexed listings (Oct 1, 2026 snapshot)
Category: Platform / Deployment
Quality Tier: 🟡 Trusted (official Vercel org, active same-day cadence; license is NOASSERTION - no recognized SPDX file)
Vercel's comprehensive ecosystem plugin packs a relational knowledge graph plus skills for every major Vercel product, specialized agents, and Vercel conventions. Unlike the narrower vercel-labs/agent-skills collection, this repo is the broad "one plugin to teach an agent the whole platform" package: AI Gateway, AI SDK, backend architecture, deployment protection, flags, queues, and a set of Vercel's own engineering-hygiene skills.
Installation¶
npx skills add vercel/vercel-plugin
Roster - Indexed Skills¶
| Skill | Installs | What It Does |
|---|---|---|
| access-protected-vercel-deployment | 2,433 | Access/test Vercel deployments behind Vercel Auth, SSO, or Deployment Protection |
| build-agents | 1,399 | Default guidance for building AI agents (eve is the default choice) |
| create-a-backend | 568 | Backend architecture: Functions, Services, containers, Workflows |
| flags-sdk | 249 | Feature flags and A/B tests with the Flags SDK + Vercel Flags |
| queues | 54 | Queue/background-processing guidance |
| custom-metrics | 35 | Custom metrics instrumentation |
| domains | 22 | Domain configuration |
| styled-jsx | 3 | styled-jsx CSS-in-JS guidance |
| ncc | 2 | ncc bundler guidance |
| edge-runtime | 2 | Edge Runtime guidance |
| streamdown | 1 | Streamdown streaming-markdown guidance |
In-repo skills not yet indexed on skills.sh (from the 43 under skills/) include: ai-gateway (setup/model discovery/routing/fallbacks/BYOK/budgets/spend), ai-sdk (AI SDK expert guidance), auth, bootstrap, cdn-caching, and more.
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Deploy + verify our sites | access-protected-vercel-deployment to test protected previews by agent |
| Build internal agents | build-agents for scaffolding a durable backend agent on Vercel |
| Frontend AI features | ai-sdk + ai-gateway for model routing with spend controls |
| Backend selection | create-a-backend to match workload to Functions/Services/Workflows |
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations / Verification¶
- skills.sh indexing verified Oct 1, 2026: 11 skills indexed; 6 SKILL.md files verified via raw.githubusercontent on branch
main(50 SKILL.md files total in-repo: 43 underskills/, 7 internal under.claude/skills/). - License is
NOASSERTION- the repo has a LICENSE file but GitHub does not recognize its SPDX identifier. Review the license text before commercial redistribution. - Distinct from the three already-documented Vercel guides (vercel-labs/agent-skills, vercel/ai, vercel/eve) - this is the broad platform-ecosystem plugin repo.
- No live install test performed; install counts and the 295-star count are from the Oct 1, 2026 sweep snapshot.
Security¶
| Audit | Verdict |
|---|---|
| Gen Agent Trust Hub | Not published |
| Socket | Not published |
| Snyk | Not published |
Related¶
← Skills Catalog | Marketplace → Powered by CorpusIQ