New update: Local Memory. Fewer tokens, faster sessions. Read the docs →
Engineering Kit · driven by /agenkitNew Local memory

Ship production software, solo.

Run /agenkit, describe what you're building, and a full team of specialist agents takes it from brainstorm to architecture, test-first build, review, and ship, pausing for your approval at each step. Production-quality thinking, without hiring the team.

One kit · seven harnesses · one license key

Your $20 AI plan dies in two projects. AgenKit makes it last.

Build any software, at any scale - solo.

In Claude Code · Cursor · Codex · +4 more →
/agenkit
  1. OrientContext Engineer
    no memory yet, indexing your codebase
  2. BrainstormConductor
    turns your idea into a clear spec
  3. ArchitectureBackend Architect
    services · data model · idempotency
  4. Build · test-firstTest Engineer
    failing test → green → refactor
  5. ReviewCode Reviewer
    spec ✓ then code quality ✓
  6. ShipInfra / DevOps
    CI/CD · deploy · verify
Running the pipeline…
Built by FAANG & frontier-AI engineers

Eleven focused agents and process skills encode production concerns: tests, security, architecture, and observability, coordinated by /agenkit.

Updated continuously

The workflows keep improving. Pull the latest anytime with one command, agenkit update, included while you're subscribed.

One simple yearly plan

A single yearly subscription covers every update. Cancel anytime; your installed kit keeps working locally.

One kit, seven harnesses, the same team everywhere

The solo-builder tax

You're doing the job of an entire team, alone.

The gap isn't effort. It's expertise. A generalist writes the code; a specialist knows the production concerns you didn't think to ask about.

What a specialist actually considers

Watch one agent reason through a routine task - the idempotency, reconciliation, indexes, and async calls a solo generalist quietly skips. That gap is what AgenKit closes, automatically.

Backend Architectthinking about: accept Stripe webhooks
  • Make it idempotent: retries deliver the same event twice; dedupe by event id.
  • Verify the signature before trusting a single byte of the payload.
  • This is I/O-bound → run it async, with a timeout, retry, and backoff.
  • Store the raw event first, ack fast, process out of band.
  • Reconcile daily against Stripe; webhooks do get dropped.

Decisions a generalist skips, applied automatically.

The /agenkit workflow

One command runs the whole engineering lifecycle.

You describe the task once. The Conductor hands each phase to the right specialist and pauses for your approval, the way a senior team ships software.

Founding price

First 150 subscribers pay $3.49/mo.

After the 150 founding seats, AgenKit is $15/mo. Claim one and your renewal never goes up.

145 of 150 founding seats left

$3.49/mothen $15/moClaim a founding seat ↗14-day refund · cancel any time
See what's inside the kit →

The full workflow, every specialist, and exactly how to use it.

Meet the eleven agents → · Workflow guides → · Install docs →

How it works

Install, activate, go.

Live in under two minutes, from checkout to a working AI team.

01

Get the CLI

Nothing to install. The commands below run with npx; Node.js 18+ required.

optional · install globally
$ npm install -g agenkit
02

Activate

Paste the license key from your receipt. The CLI stores it locally.

activate
$ npx agenkit activate <license-key>
03

Install the kit

Run it in your project folder. It auto-detects your harness and installs the right files.

install the kit
$ npx agenkit install engineering-kit

Any stack

Works in any language, any architecture.

Install it into any project, in any language. The agents read your existing code and follow your conventions, from a single service to a modular monolith to microservices.

Why builders keep it installed

⌘ Seven harnesses, one kit

Claude Code, Cursor, Codex, OpenCode, Antigravity, GitHub Copilot, and Kimi Code - the same team everywhere.

↓ One command install

agenkit install auto-detects your harness and drops files where each tool expects them.

✓ Checksum-verified

Every artifact is SHA-256 verified before it touches your project. No silent corruption.

⌥ Driven by /agenkit

One slash command runs the whole pipeline - brainstorm, architecture, plan, test-first build, review, ship.

⟳ Update in place

agenkit update pulls the latest versions. Your own files are never overwritten.

≈ Any stack

Agents adapt to your codebase and conventions instead of imposing an architecture.

Learn

Docs, guides, and specialist agents.

Install with the docs, sharpen your workflow with guides, and see who runs each /agenkit phase.

Comparison

Why builders choose AgenKit.

Other kitsAgenKit
Harness supportMany kits target one tool✓All seven harnesses, one license
InstallManual copy or per-tool scripts✓npx agenkit install, auto-detects harness
What you getA small agent pack✓Eleven specialists plus /agenkit workflow
Pricing modelOften one-time or per-seat SaaS✓One yearly subscription, all harnesses
Checksum-verified installsUncommon✓SHA-256 on every download
Updates over timeVaries by project✓agenkit update while subscribed

Social proof

Builders are already shipping with AgenKit.

Trusted by professional and senior developers across all seven harnesses.

Pricing

One kit. One yearly subscription.

The subscription pays for the updates: the kit keeps evolving, and every new release is included, across all seven harnesses. No seats, no account to manage.

Full pricing & FAQ →

FAQ

Questions, answered.

Still curious? Everything installs locally and keeps working offline once installed.

01 Which tools does it work with?+
All seven: Claude Code, Cursor, Codex CLI, OpenCode, Antigravity CLI, GitHub Copilot, and Kimi Code. The kit is compiled to the exact file layout each tool expects.
02 How do I start it in my tool?+
Trigger AgenKit with /agenkit in most tools (Claude Code, Cursor, OpenCode, Antigravity, and Kimi Code). In GitHub Copilot it's @agenkit, and in Codex you invoke the AgenKit skill by name. The install step prints the exact trigger for your harness.
03 What exactly do I get?+
The Engineering Kit - a full team of specialist agents, the process skills that guide them, and the /agenkit command - delivered as versioned archives installed by the agenkit CLI. Your license key is the only credential; there is no account to manage.
04 Do I need to create an account?+
No. Checkout issues a license key via Polar; the CLI stores it locally and the API validates it on install.
05 Is it one-time or a subscription?+
A simple subscription that works out to just $3.49/mo (billed yearly). Early subscribers keep that founding rate while subscribed; after that it's $15/mo. It covers every update while active, across all seven harnesses. Cancel any time; the installed kit keeps working locally.
06 How do updates work?+
Run agenkit update. It compares your lockfile to the latest versions and refreshes everything your subscription unlocks.
07 Is there a refund policy?+
Yes - a 14-day money-back window, handled through Polar's customer portal.
08 Is AgenKit related to OpenAI's AgentKit?+
No. AgenKit (agenkit.xyz) is an independent product: a guided engineering team you install into Claude Code, Cursor, Codex, and other harnesses. It is not affiliated with OpenAI AgentKit.

Stop being the bottleneck. Ship like a team.

A full engineering team in every harness you use, driven by one command: /agenkit.

Subscribe · $3.49/mo ↗