vercel-cost-tuning — Claude Code skill for cutting Vercel spend
vercel-cost-tuning reads Vercel billing and finds where money leaks: function time, bandwidth, middleware calls. Use it before the next invoice.
MIT 1 installs ★ 2,935 updated 2026-11-01
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a claude-codeWhat vercel-cost-tuning does
The skill walks through Fluid Compute pricing and digs into the main cost drivers. Function CPU and memory, bandwidth from large responses, edge middleware that fires on every request, image transforms, and build minutes. It pulls usage via the Vercel API, then suggests changes in vercel.json and route code.
One path it takes: right-size memory per endpoint, so a lightweight JSON route gets 128 MB instead of the default. Another: move read-only handlers to Edge Functions and add s-maxage cache headers so cached routes never hit the function at all.
You need access to the Vercel billing dashboard and a $VERCEL_TOKEN for API calls. The skill also covers spend management: set a monthly budget, enable a hard stop, get email alerts at 50%, 75%, 90%, 100%. Middleware is a common surprise, since it runs on every matched request, so the skill scopes it with a matcher.
What it helps with
- Pull current Vercel usage via API
- Cut serverless execution time by right-sizing memory
- Add edge caching so cached routes skip functions
- Narrow middleware matcher to relevant paths
- Set spend management budget and alerts
Example prompts for the agent
- Analyze my Vercel billing and list the biggest cost drivers
- Right-size function memory in vercel.json for my API routes
- Set up spend management alerts at 75% and 100%
How to install vercel-cost-tuning
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a claude-code2Manually: put the files in place
~/.claude/skills/vercel-cost-tuning/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/vercel-cost-tuning/
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a codex2Manually: put the files in place
.agents/skills/vercel-cost-tuning/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a cursor2Manually: put the files in place
.cursor/skills/vercel-cost-tuning/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install jeremylongshore/tons-of-skills-marketplace skills/.curated/vercel-cost-tuning2Manually: put the files in place
.github/skills/vercel-cost-tuning/SKILL.md1Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a gemini-cli2Platform CLI
gemini skills install https://github.com/jeremylongshore/tons-of-skills-marketplace --path skills/.curated/vercel-cost-tuningManually: put the files in place
.devin/skills/vercel-cost-tuning/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -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-cost-tuning
How do I install vercel-cost-tuning in Claude Code?
Run npx skills add jeremylongshore/tons-of-skills-marketplace --skill vercel-cost-tuning -a claude-code, then check ~/.claude/skills/vercel-cost-tuning/SKILL.md.
What does vercel-cost-tuning do?
It works through Vercel Fluid Compute pricing and finds reductions in function time, bandwidth, middleware invocations, image optimizations, and build minutes.
What do I need before running it?
Access to the Vercel billing dashboard and a Vercel API token. Knowing your deployment architecture helps, along with access to Vercel Analytics.
Source text
SKILL.md · MIT
--- name: vercel-cost-tuning description: 'Optimize Vercel costs through plan selection, function efficiency, and usage monitoring. Use when analyzing Vercel billing, reducing function execution costs, or implementing spend management and budget alerts. Trigger with phrases like "vercel cost", "vercel billing", "reduce vercel costs", "vercel pricing", "vercel expensive", "vercel budget". ' allowed-tools: Read, Write, Edit, Bash(vercel:*), Bash(curl:*), Grep version: 1.18.0 license: MIT author: Jeremy Longshore <[email protected]> tags: - saas - vercel - cost-optimization - billing - monitoring compatibility: Designed for Claude Code --- # Vercel Cost Tuning ## Overview Optimize Vercel costs by understanding the Fluid Compute pricing model, reducing function execution time, leveraging edge caching to avoid function invocations, and configuring spend management. Covers plan comparison, cost drivers, and monitoring. ## Prerequisites - Access to Vercel billing dashboard - Understanding of current deployment architecture - Access to Vercel Analytics for usage patterns ## Instructions ### Step 1: Understand the Pricing Model Vercel uses **Fluid Compute** pricing (for new projects): | Resource | Hobby (Free) | Pro ($20/member/mo) | Enterprise | |----------|-------------|---------------------|------------| | Bandwidth | 100 GB | 1 TB included | Custom | | Serverless Execution | 100 GB-hrs | 1000 GB-hrs included | Custom | | Edge Function invocations | 500K | 1M included | Custom | | Edge Middleware invocations | 1M | 1M included | Custom | | Image Optimizations | 1000 | 5000 included | Custom | | Builds per day | 6000 | 6000 | Custom | | Concurrent builds | 1 | 1 (more available) | Custom | **Fluid Compute billing breakdown:** - **Active CPU time**: charged per ms of actual CPU usage - **Provisioned memory**: charged per GB-second of allocated memory - Benefit: you pay for actual work, not idle waiting (e.g., waiting for a database response) ### Step 2: Identify Cost Drivers ```bash # Check usage via API curl -s -H "Authorization: Bearer $VERCEL_TOKEN" \ "https://api.vercel.com/v2/usage" | jq . # Top cost drivers on Vercel: # 1. Serverless function execution time (CPU + memory) # 2. Bandwidth (large responses, unoptimized images) # 3. Edge Middleware invocations (runs on EVERY request) # 4. Image optimizations (each unique transform costs) # 5. Build minutes (frequent deploys or slow builds) ``` ### Step 3: Reduce Function Execution Costs ```typescript // 1. Right-size function memory — don't over-allocate // vercel.json { "functions": { "api/lightweight.ts": { "memory": 128 }, // Simple JSON responses "api/standard.ts": { "memory": 512 }, // Database queries "api/heavy.ts": { "memory": 1024 } // Image processing } } // 2. Move read-only endpoints to Edge Functions (cheaper, no cold starts) // api/config.ts export const config = { runtime: 'edge' }; export default function handler() { return Response.json({ features: ['a', 'b'] }); } // 3. Cache function responses at the edge // Eliminates function invocations entirely for cached routes export default function handler(req, res) { res.setHeader('Cache-Control', 's-maxage=3600, stale-while-revalidate=86400'); res.json(data); } ``` ### Step 4: Reduce Bandwidth Costs ```json // vercel.json — compress and cache aggressively { "headers": [ { "source": "/static/(.*)", "headers": [ { "key": "Cache-Control", "value": "public, max-age=31536000, immutable" } ] } ], "images": { "sizes": [640, 750, 1080], "formats": ["image/avif", "image/webp"], "minimumCacheTTL": 86400 } } ``` Key bandwidth reducers: - Use Vercel's image optimization (auto WebP/AVIF conversion) - Set aggressive cache headers on static assets - Use ISR to serve static HTML instead of SSR - Compress API responses (Vercel auto-compresses with Brotli) ### Step 5: Optimize Middleware Costs Middleware runs on **every matched request**. Minimize its scope: ```typescript // middleware.ts — scope to specific paths only export const config = { matcher: [ // Only run middleware on API routes and protected pages '/api/:path*', '/dashboard/:path*', // Skip static files, images, and public assets '/((?!_next/static|_next/image|favicon.ico|public).*)', ], }; export function middleware(request) { // Keep logic minimal — this runs on every matched request // Avoid: database queries, external API calls, heavy computation // Good: cookie checks, header modifications, redirects } ``` ### Step 6: Configure Spend Management In the Vercel dashboard under **Settings > Billing > Spend Management**: ``` Default budget: $200/month on-demand usage Options: - Set custom budget limit - Enable hard limit (pauses all projects when reached) - Configure email alerts at 50%, 75%, 90%, 100% ``` ```bash # Check current usage against budget via API curl -s -H "Authorization: Bearer $VERCEL_TOKEN" \ "https://api.vercel.com/v2/usage?teamId=team_xxx" \ | jq '{period: .period, bandwidth: .bandwidth, execution: .serverlessFunctionExecution}' ``` ## Cost Optimization Checklist | Action | Impact | Effort | |--------|--------|--------| | Add `s-maxage` cache headers | High — eliminates function invocations | Low | | Use Edge Functions for simple endpoints | Medium — cheaper than serverless | Low | | Right-size function memory | Medium — reduces GB-hr cost | Low | | Scope middleware matcher | Medium — reduces edge invocations | Low | | Enable image optimization | Medium — reduces bandwidth | Low | | Use ISR instead of SSR | High — serves cached HTML | Medium | | Optimize build speed | Low — reduces build minutes | Medium | | Set spend management alerts | Safety — prevents surprise bills | Low | ## Output - Function memory right-sized per endpoint - Edge caching reducing function invocations - Middleware scoped to minimize invocations - Spend management configured with budget alerts - Usage monitoring via API ## Error Handling | Error | Cause | Solution | |-------|-------|----------| | Unexpected bill spike | Uncached high-traffic endpoint | Add `s-maxage` to the response | | Projects paused | Hard spending limit reached | Increase limit or optimize usage | | Image optimization quota exceeded | Too many unique image transforms | Reduce `sizes` array, increase cache TTL | | Build minutes exceeded | Slow builds or too many deploys | Use `ignoreCommand` to skip non-code changes | ## Examples ### Reduce an invocation spike with a measurable rollback For a high-traffic read endpoint, add a conservative `s-maxage` response policy in a preview deployment and compare invocation count, cache hit rate, and response correctness against the prior release using synthetic requests. Set spend alerts before rollout and avoid a hard account-wide pause unless the owner accepts its blast radius. If personalized content is accidentally cached, immediately roll back the header change, purge only the affected deployment cache where supported, and document the regression. ## Resources - [Vercel Pricing](https://vercel.com/pricing) - [Fluid Compute Pricing](https://vercel.com/docs/functions/usage-and-pricing) - [Spend Management](https://vercel.com/docs/pricing#spend-management) - [Limits](https://vercel.com/docs/limits) - [Usage API](https://vercel.com/docs/rest-api) ## Next Steps For reference architecture, see `vercel-reference-architecture`.
© Jeremy Longshore · MIT · jeremylongshore/tons-of-skills-marketplace