vercel-policy-guardrails — Claude Code skill
The vercel-policy-guardrails skill adds Vercel project checks for secret exposure, CI policy violations, Edge Runtime imports, and protected routes.
MIT 1 installs ★ 2,935 updated 2026-11-01
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a claude-codeWhat vercel-policy-guardrails does
The skill sets guardrails for Vercel projects. It checks sensitive variables that use the NEXT_PUBLIC_ prefix, blocks Node.js imports in Edge Runtime files, and scans staged files for credential patterns before a commit.
Its setup includes ESLint rules, a pre-commit hook, CI validation for vercel.json, Edge Runtime checks, a .next/ output-size check, and a guard that keeps .env.example aligned with variables used in code. It also includes middleware for enforcing authentication on protected routes. The project needs ESLint, Git hook infrastructure, a CI/CD pipeline, and TypeScript. Without ESLint configured, the main lint rules cannot run.
What it helps with
- block secrets from client bundles
- scan staged files before commits
- validate
vercel.jsonin CI - catch Node.js imports in Edge Runtime code
- keep
.env.examplealigned with code
Example prompts for the agent
- Set up Vercel guardrails that prevent secrets from reaching client bundles.
- Add CI checks for
vercel.json, Edge Runtime imports, and the build output size. - Check that protected routes require authentication and every environment variable is documented in
.env.example.
How to install vercel-policy-guardrails
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a claude-code2Manually: put the files in place
~/.claude/skills/vercel-policy-guardrails/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/vercel-policy-guardrails/
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a codex2Manually: put the files in place
.agents/skills/vercel-policy-guardrails/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a cursor2Manually: put the files in place
.cursor/skills/vercel-policy-guardrails/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install jeremylongshore/tons-of-skills-marketplace skills/.curated/vercel-policy-guardrails2Manually: put the files in place
.github/skills/vercel-policy-guardrails/SKILL.md1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a gemini-cli2Platform CLI
gemini skills install https://github.com/jeremylongshore/tons-of-skills-marketplace --path skills/.curated/vercel-policy-guardrailsManually: put the files in place
.devin/skills/vercel-policy-guardrails/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about vercel-policy-guardrails
How do I install vercel-policy-guardrails in Claude Code?
Run npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-policy-guardrails -a claude-code, then check ~/.claude/skills/vercel-policy-guardrails/SKILL.md.
What does vercel-policy-guardrails do?
It adds ESLint rules, credential scanning before commits, and CI policy checks for Vercel projects. The checks cover Edge Runtime compatibility, build output size, environment variable documentation, and route authentication.
What does the project need before using this skill?
The project needs configured ESLint, Git hook infrastructure, a CI/CD pipeline, and TypeScript. The checks work with files such as vercel.json and .env.example.
Source text
SKILL.md · MIT
--- name: vercel-policy-guardrails description: 'Implement lint rules, CI policy checks, and automated guardrails for Vercel projects. Use when setting up code quality rules, preventing secret exposure, or enforcing deployment policies for Vercel applications. Trigger with phrases like "vercel policy", "vercel lint", "vercel guardrails", "vercel best practices check", "vercel secret scan". ' allowed-tools: Read, Write, Edit, Bash(npx:*), Bash(npm:*) version: 1.18.0 license: MIT author: Jeremy Longshore <[email protected]> tags: - saas - vercel - policy - linting - security compatibility: Designed for Claude Code --- # Vercel Policy Guardrails ## Overview Protect against common Vercel failure modes with automated guardrails: ESLint rules preventing secret exposure in client bundles, pre-commit hooks scanning for credentials, CI checks validating vercel.json and edge runtime compatibility, and runtime middleware enforcing auth on protected routes. ## Prerequisites - ESLint configured in project - Git hooks infrastructure (husky or lefthook) - CI/CD pipeline (GitHub Actions or similar) - TypeScript for type enforcement ## Instructions ### Step 1: ESLint Rules — Prevent Secret Exposure ```javascript // .eslintrc.js — custom rules for Vercel projects module.exports = { rules: { // Prevent using NEXT_PUBLIC_ prefix for sensitive variables 'no-restricted-syntax': [ 'error', { selector: 'MemberExpression[object.property.name="env"][property.name=/^NEXT_PUBLIC_(SECRET|KEY|TOKEN|PASSWORD|PRIVATE)/]', message: 'Do not prefix secrets with NEXT_PUBLIC_ — they will be exposed in the client bundle', }, ], }, overrides: [ { // Edge runtime files — prevent Node.js API usage files: ['**/edge-*.ts', '**/middleware.ts'], rules: { 'no-restricted-imports': [ 'error', { paths: [ { name: 'fs', message: 'fs is not available in Edge Runtime. Use fetch or Vercel Blob.' }, { name: 'path', message: 'path is not available in Edge Runtime. Use URL API.' }, { name: 'crypto', message: 'Use globalThis.crypto (Web Crypto API) in Edge Runtime.' }, { name: 'child_process', message: 'child_process is not available in Edge Runtime.' }, { name: 'net', message: 'net is not available in Edge Runtime.' }, ], }, ], }, }, ], }; ``` ### Step 2: Pre-Commit Hook — Credential Scanning ```bash # Install husky npm install --save-dev husky npx husky init ``` ```bash #!/usr/bin/env bash # .husky/pre-commit set -euo pipefail # Scan staged files for credentials PATTERNS=( 'VERCEL_TOKEN\s*[:=]\s*\S+' 'vercel_[a-zA-Z]*_token\s*[:=]\s*\S+' 'sk_live_[a-zA-Z0-9]+' 'NEXT_PUBLIC_.*SECRET' 'api\.vercel\.com.*Bearer\s+[a-zA-Z0-9]+' ) STAGED_FILES=$(git diff --cached --name-only --diff-filter=ACM) FOUND=0 for file in $STAGED_FILES; do for pattern in "${PATTERNS[@]}"; do if grep -qEi "$pattern" "$file" 2>/dev/null; then echo "ERROR: Potential credential found in $file" echo " Pattern: $pattern" FOUND=1 fi done done if [ $FOUND -ne 0 ]; then echo "" echo "Commit blocked: Remove credentials and use environment variables." echo "See: vercel env add <KEY> <environment>" exit 1 fi ``` ### Step 3: CI Policy Check — vercel.json Validation ```yaml # .github/workflows/vercel-policy.yml name: Vercel Policy Checks on: [pull_request] jobs: policy: runs-on: ubuntu-latest steps: - uses: actions/checkout@v4 - uses: actions/setup-node@v4 with: { node-version: 20 } - name: Validate vercel.json schema run: | if [ -f vercel.json ]; then node -e " const config = require('./vercel.json'); const errors = []; // Check for deprecated builds property if (config.builds) { errors.push('vercel.json uses deprecated \"builds\" property — use \"functions\" instead'); } // Check compressHTML (iOS Safari issue) if (config.compressHTML === true) { errors.push('compressHTML should be disabled — causes iOS Safari rendering issues'); } // Check for hardcoded secrets in headers const headerStr = JSON.stringify(config.headers ?? []); if (/Bearer\s+[a-zA-Z0-9]{20,}/.test(headerStr)) { errors.push('Hardcoded token found in vercel.json headers'); } if (errors.length > 0) { console.error('Policy violations:'); errors.forEach(e => console.error(' - ' + e)); process.exit(1); } console.log('vercel.json policy checks passed'); " fi - name: Check edge runtime compatibility run: | # Find files with edge runtime declaration for file in $(grep -rl "runtime.*=.*'edge'" src/ api/ 2>/dev/null || true); do echo "Checking edge compatibility: $file" # Check for Node.js-only imports if grep -E "require\(['\"]fs['\"]|from ['\"]fs['\"]|from ['\"]path['\"]|from ['\"]crypto['\"]" "$file"; then echo "ERROR: $file uses Node.js APIs incompatible with Edge Runtime" exit 1 fi done echo "Edge runtime compatibility checks passed" - name: Check bundle size budget run: | npm ci npm run build # Check output size TOTAL=$(du -sb .next/ 2>/dev/null | cut -f1 || echo 0) MAX=$((250 * 1024 * 1024)) # 250MB if [ "$TOTAL" -gt "$MAX" ]; then echo "ERROR: Build output ($TOTAL bytes) exceeds budget ($MAX bytes)" exit 1 fi echo "Bundle size within budget: $TOTAL bytes" ``` ### Step 4: Env Var Documentation Guard ```bash #!/usr/bin/env bash # scripts/check-env-docs.sh — ensure .env.example stays in sync set -euo pipefail # Extract env vars used in code CODE_VARS=$(grep -roh 'process\.env\.\w\+' src/ api/ 2>/dev/null \ | sed 's/process\.env\.//' \ | sort -u) # Extract vars documented in .env.example if [ ! -f .env.example ]; then echo "ERROR: .env.example file missing" exit 1 fi DOC_VARS=$(grep -oE '^\w+=' .env.example | sed 's/=//' | sort -u) # Find undocumented vars MISSING=$(comm -23 <(echo "$CODE_VARS") <(echo "$DOC_VARS")) if [ -n "$MISSING" ]; then echo "ERROR: Undocumented environment variables:" echo "$MISSING" | sed 's/^/ /' echo "Add these to .env.example" exit 1 fi echo "All environment variables documented" ``` ### Step 5: Runtime Auth Middleware Guard ```typescript // middleware.ts — enforce that protected routes always require auth import { NextRequest, NextResponse } from 'next/server'; // Routes that MUST require authentication const PROTECTED_PATTERNS = [ /^\/api\/admin/, /^\/api\/users/, /^\/dashboard/, /^\/settings/, ]; // Routes explicitly allowed without auth const PUBLIC_PATTERNS = [ /^\/api\/health/, /^\/api\/webhooks/, /^\/$/, // homepage /^\/login/, /^\/signup/, ]; export function middleware(request: NextRequest) { const { pathname } = request.nextUrl; const isProtected = PROTECTED_PATTERNS.some(p => p.test(pathname)); const isPublic = PUBLIC_PATTERNS.some(p => p.test(pathname)); if (isProtected && !isPublic) { const token = request.cookies.get('session')?.value; if (!token) { if (pathname.startsWith('/api/')) { return NextResponse.json({ error: 'Unauthorized' }, { status: 401 }); } return NextResponse.redirect(new URL('/login', request.url)); } } return NextResponse.next(); } export const config = { matcher: ['/((?!_next/static|_next/image|favicon.ico).*)'], }; ``` ### Step 6: Deployment Freeze Guard ```bash #!/usr/bin/env bash # scripts/check-deploy-freeze.sh — prevent production deploys during freeze windows set -euo pipefail # Check if we're in a deployment freeze window HOUR=$(date -u +%H) DAY=$(date -u +%u) # 1=Monday, 7=Sunday # No deploys on weekends if [ "$DAY" -gt 5 ]; then echo "BLOCKED: No production deploys on weekends" exit 1 fi # No deploys after 4pm UTC (Friday especially) if [ "$DAY" -eq 5 ] && [ "$HOUR" -ge 16 ]; then echo "BLOCKED: No production deploys after 4pm UTC on Fridays" exit 1 fi echo "Deploy allowed" ``` ## Guardrails Summary | Guardrail | Enforcement Point | Prevents | |-----------|-------------------|----------| | Secret prefix lint | ESLint (editor + CI) | Client bundle secret exposure | | Edge runtime lint | ESLint (editor + CI) | Node.js APIs in edge functions | | Credential scan | Pre-commit hook | Secrets in version control | | vercel.json validation | CI | Deprecated config, hardcoded tokens | | Env var documentation | CI | Missing .env.example entries | | Auth middleware | Runtime | Unprotected routes | | Deploy freeze | CI | Weekend/late deploys | ## Output - ESLint rules preventing secret exposure and edge runtime violations - Pre-commit hooks blocking credentials from entering git - CI policy checks validating configuration and compatibility - Runtime middleware enforcing authentication on protected routes - Deployment freeze windows preventing risky deploys ## Error Handling | Error | Cause | Solution | |-------|-------|----------| | ESLint rule false positive | Variable name matches pattern | Add `// eslint-disable-next-line` with justification | | Pre-commit hook blocks valid commit | Pattern too broad | Narrow the regex or add allowlist | | CI edge check false positive | Dead import | Remove unused import | | Deploy freeze too restrictive | Urgent hotfix needed | Use `--force` flag with team approval | ## Examples ### Authorize an urgent deployment-freeze exception For a production incident, create a time-boxed exception that names the incident, deployment SHA, approving owner, and rollback deployment. Run required tests and a synthetic preview check before the exception is used; the bypass must be scoped to this release rather than disabling the freeze globally. After recovery, revoke the exception, review why the normal window could not be used, and add a focused prevention item to the incident record. ## Resources - [ESLint Custom Rules](https://eslint.org/docs/latest/extend/plugins) - [Husky Documentation](https://typicode.github.io/husky/) - [Vercel Project Configuration](https://vercel.com/docs/project-configuration) - [Edge Runtime API](https://vercel.com/docs/functions/runtimes/edge) ## Next Steps For architecture variants, see `vercel-architecture-variants`.
© Jeremy Longshore · MIT · jeremylongshore/tons-of-skills-marketplace