Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add augchan42/inkstone --skill vercel-build-speedgit clone --depth 1 https://github.com/augchan42/inkstoneWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/augchan42/inkstone/vercel-build-speed)<a href="https://agentmods.dev/skills/augchan42/inkstone/vercel-build-speed"><img src="https://agentmods.dev/badge/skills/augchan42/inkstone/vercel-build-speed/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/augchan42/inkstone/vercel-build-speed"><img src="https://agentmods.dev/badge/skills/augchan42/inkstone/vercel-build-speed.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00064 | $0.02791 |
| Opus 5 | $0.00032 | $0.01396 |
| Sonnet 5 | $0.00013 | $0.00558 |
| Haiku 4.5 | $0.00006 | $0.00279 |
Grade C, and why
vercel-build-speed scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf .next && npm run build # compare "Compiled successfully in Xs" How it starts
The opening of the file, as written. The whole thing — 254 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Vercel Build Speed (Next.js App Router)
Derived from taking a real production Next.js 16 / Turbopack app from 7m21s to 3m46s, measured per-phase on Vercel at every step. The single most useful lesson was epistemic: the obvious lever was the wrong one. Removing ~560MB of assets bought about 5 seconds. Cutting prerendered page count bought ~50.
Calibration — this is one case study, not a benchmark
Every number here comes from one app, one Vercel plan, one 4-core / 8GB build
machine, in mid-2026 on Next 16.2. The app was a localized content site:
four locales, ~2500 prerendered pages, a large public/, a 1.5GB git history.
Weight the findings accordingly:
Likely to generalize — the method (read the phase table, change one class of
thing per deploy, re-attribute after each), the vercel.json failure mode, the
Turbopack dead-config audit, and the shape of the ranking: compute and page
count beat bytes.
Re-measure for your project — every specific duration, and especially the
three "myths". "Clone time is not proportional to tree size" and "Deploying
outputs has a fixed floor" each rest on two observations on one project.
They may be artifacts of this repo's history size or this plan's build
infrastructure. If your public/ dominates a small output, or your history is
shallow, you may see the opposite. Do not skip a change because this document
says the lever is weak — measure it.
Has a shelf life — Turbopack and the Next config surface move fast. The
experimental.bundleAnalyzer note in particular is a snapshot of 16.2 behavior
and will likely stop being true. Verify config keys against your own build
output, never against this file.
Rule 0: Measure the phases before changing anything
Every Vercel build log is already an itemised profile. Pull it and build the table before forming any theory.
Cloning completed: 21.6s
✓ Compiled successfully in 31.8s
Finished TypeScript in 29.3s
✓ Generating static pages (2549/2549) in 84s
Build Completed in /vercel/output [4m]
Deploying outputs... ← next timestamp minus this one
Created build cache: 20s
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 254 lines · 64 tokens per session scan C b12b17a1a396
vercel-build-speed is a skill published in the GitHub repository augchan42/inkstone (4 stars, last pushed 12d ago), licensed MIT. It adds 64 tokens to every session and 2,791 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
deploy-site
Deploys an existing Power Pages code site to a Power Pages environment using PAC CLI. Handles tooling verification, authentication, environment confirmation, building, and uploading. Use when the user wants to deploy, upload, or publish their code site.
activate-site
Activates and provisions a Power Pages website in a Power Platform environment via the Power Platform REST API. Use when the user wants to activate, provision, turn on, or enable a Power Pages website or portal.
deploy
Builds and deploys a Power Apps code app to Power Platform. Use when deploying changes, redeploying an existing app, or pushing updates.
domain-monitoring
AI-powered domain and SSL certificate monitoring skill. Designs domain health checks, SSL expiry alerts, DNS anomaly detection, and security rating assessments for e-commerce websites.
public-status-page
AI-powered public status page design skill for e-commerce businesses. Creates status page architecture, incident communication templates, subscriber notification systems, and uptime reporting frameworks.
deploy
Render Quarto .qmd slides to HTML and sync to docs/ for GitHub Pages. Use when user says "deploy", "publish the slides", "ship to pages", "push the lecture live", "render and publish", or after Quarto edits that need to go public. NOT for local Quarto render only — use quarto render directly for that.