Is this a one-time payment or a subscription?

One-time. Buy a kit or the Complete Bundle and it's yours forever — no seat fees, no annual renewals, no surprises.

What exactly do I get when I buy a kit?

Full source code, a CLAUDE.md, setup guide, architecture notes, and a regression eval suite. Everything wired together and ready to run — not a skeleton you still have to build.

Can I use these in commercial / client projects?

Yes. Every purchase includes a commercial license. The Team license adds multi-seat usage and unlimited client projects.

What happens when a new model is released?

We re-run every eval suite against the new model, fix whatever breaks, and push the update to all owners — free, with a changelog. You never have to chase a model upgrade yourself.

Do you offer refunds?

14-day money-back guarantee, no questions asked. Email us, we refund. Done.

Are these actual working kits or just templates?

Working kits. Each one ships with a passing eval suite — you can see the results before you buy. We don't sell starters you still have to finish; we sell production-ready codebases you customize.

What if Claude Code changes and the kits break?

That's our problem, not yours. When a new Claude Code version ships, we update the affected kits and push the fix to every owner. The eval suite tells us immediately if something regresses.

I'm a solo developer. Is this overkill?

It's the opposite. Solo developers get the most value — you skip weeks of trial-and-error scaffolding and ship something a team would be proud of. The kits are designed to be maintained by one person.

5 battle-tested kits · one-time payment

Ship agents that
actually work.

Five battle-tested Claude Code & AI agent kits — done right, not done fast. Each ships with docs, evals and lifetime updates. Buy one, or own the whole set.

Browse the kits See how it works

9,200+ builders shipping with Agent Cookbook

~/projects/acme

# scaffold a production agent in one command

$ npx agentcookbook add support-agent-workbench

✓ fetching template · verified build

✓ wiring CLAUDE.md, commands, hooks

✓ installing eval suite (42 cases)

→ ready in 8.4s. run npm run agent

Evals passing

42 / 42 · 100%

5 battle-tested kits · one-time payment

Ship agents that actually work.

Five battle-tested Claude Code & AI agent kits — done right, not done fast. Docs, evals and lifetime updates included. Skip the boilerplate, keep the quality.

Browse the kits See how it works

~/projects/acme

# scaffold a production agent in one command

$ npx agentcookbook add support-agent-workbench

✓ wiring CLAUDE.md, commands, hooks, eval suite

→ ready in 8.4s

Powering builders & teams everywhere

NorthwindCobaltHelioMarlow&CoDriftQuanta

Why Agent Cookbook

Most agent templates rot in a week. Ours don't.

We'd rather sell five kits we'd stake our name on than a hundred we wouldn't. Each one is reviewed, tested, and re-versioned the day a new model drops.

Battle-tested in production

Every kit runs in a real app before it's listed. No toy demos, no “works on my machine.”

Docs + eval suites included

Setup guides, architecture notes, and a regression eval harness so you know it works — and stays working.

Lifetime updates, free

Buy once. When a new model ships, we re-tune and re-test — and you get the update at no extra cost.

Flagship kits

9.2k

Active builders

4.9★

Avg. rating

∞

Lifetime updates

The lineup

Five kits. Zero fluff.

Browse all five

Getting StartedClaude Code

Claude Code Project Bootstrap

Opinionated CLAUDE.md, slash-commands and hooks that make Claude Code productive in any repo.

★ 5.0 · 2 reviews$39

Starter KitNext.js

Agent-Ready SaaS Starter

Auth, billing, and a wired-in Claude agent layer. Launch a SaaS with agents on day one.

★ 5.0 · 2 reviews$129

IntegrationClaude Code

Legacy Repo Integration Kit

Safely add Claude Code agents to an existing codebase — guardrails, conventions, incremental adoption.

★ 4.5 · 2 reviews$89

MarketingNext.js

SEO Marketing Site + Content Agent

A fast, SEO-first Next.js site plus an agent that researches, drafts, and publishes content.

★ 5.0 · 2 reviews$99

App TemplateNext.js

Support Agent Workbench

A production triage + reply agent with tools, guardrails, and a human-in-the-loop console. Includes a 42-case eval suite.

★ 5.0 · 2 reviews$129

Best value

Want all five?
Get the Complete Bundle.

Own every kit + every future release. One payment, save $136.

See the bundle

Free starter

Try the quality before you buy

Grab a trimmed Claude Code starter — a clean CLAUDE.md plus 8 of our most-used commands. Free, no catch.

Join 12,000+ builders. No spam — unsubscribe anytime.

starter-kit/

CLAUDE.md

commands/

/review /test /ship

/scaffold /refactor

README.md

From zero to shipped

Three steps. Minutes, not weeks.

Pick your kit

Browse the five. Preview the file tree, eval results and live demo before you buy.

Install in one command

Run npx agentcookbook add — it wires config, commands and dependencies automatically.

Customize & ship

Clear extension points and docs. Bend it to your product, run the evals, and deploy with confidence.

“

Agent Cookbook saved us a full sprint. The support-agent kit was in production by Thursday — evals and all.

Priya Raman

Founding Engineer, Cobalt

Pricing

Pay once. Own it forever.

No subscriptions. Buy a single kit, or grab the Complete Bundle once and own everything — including the next kits we ship.

Single kit

For one specific job.

$39+

Browse the kits

✓ One-time payment
✓ Lifetime updates for that kit
✓ Docs + eval suite
✓ Commercial license

Best value · save $136

Complete Bundle

All five kits, one payment.

$349$485

✓ All 5 flagship kits
✓ Every future kit, free
✓ Lifetime updates on everything
✓ Priority support + private Discord

Team license

For agencies & squads.

$699

Talk to us

✓ Everything in the Bundle
✓ Up to 10 seats
✓ Use across unlimited client projects
✓ Onboarding workshop

Every purchase is a one-time payment with a 14-day money-back guarantee.

From the blog

Guides for building with agents

All articles

CLAUDE CODE · 7 MIN

Writing a CLAUDE.md that actually steers the agent

The structure, the anti-patterns, and a copy-paste starting template.

EVALS · 11 MIN

How to eval an agent before it ever sees a user

Building a regression harness that catches model drift early.

GROWTH · 6 MIN

Shipping an SEO content pipeline with one agent

From keyword to published post, fully automated and reviewed.

FAQ

Questions, answered

Stop scaffolding. Start shipping.

Join 9,200+ builders using Agent Cookbook to launch agents that survive production.

Browse the kits Get the bundle

Ship agents thatactually work.