ShipTrack
Sign inStart for free

For people building with Claude Code, Codex, Antigravity and Cursor

Track the SaaS your AI agent is building, from idea to first sale.

ShipTrack turns your idea into versions and steps, has the coding agent on your own computer do them, and shows live whether each step is really done. After launch, your revenue lives in the same place.

Start for free See how it works
  • No credit card needed
  • Your code stays on your computer
  • English and Turkish
shiptrack.ensatech.dev / studioflow

Now: v1

8/18 steps
MVP 6/6 v1 2/6 Launch 0/6
ST-7Packages and memberships
ST-8Deduct credits on booking
ST-9Buy packages online (Stripe)
ST-10Student management and CRM
ST-11Notifications and reminders
ST-12Mobile-friendly student experience

How it works

One idea, nine stations.

The belt shows a project's real journey through ShipTrack. At every station a worker adds something to your idea; click a station to read what ShipTrack does there.

01/09You describe your idea

One prompt is enough: say what you want to build and who it's for. ShipTrack hands it to Claude.

Canvas

Your whole project on one screen.

Versions line up in lanes, steps sit with the dependency arrows between them. As your agent works the cards turn green live; UIForge screens and launch prep live on the same canvas.

  • Arrows show which step waits for what
  • See live which step your agent is on
  • Design screens and launch tasks at a glance
studioflow / canvas

Why

Getting AI to write code got easy. Keeping track of it didn't.

The plan gets lost in chat history.

Your idea becomes versions, steps, acceptance criteria and a ready-made agent prompt for every step. The plan stays, and you can edit it.

You don't know what the agent did, or whether it's really done.

When the agent finishes a step, it turns green on its own. You see, line by line, which files changed and how.

Keys, launch and revenue all live in other tabs.

Missing API keys, launch status and your Stripe or Lemon Squeezy revenue sit inside the same project.

01 · Plan

From a one-paragraph idea to a roadmap you can execute

Describe what you want to build in your own words. Claude works out the core idea, audience and value proposition, then writes the versions from MVP to launch and the steps inside each one.

  • Every step says what to do, when it counts as done, and what to tell the agent
  • Reorder, edit, delete or postpone steps whenever you like
  • The plan is written live on screen and is usually ready within two minutes
A subscription web app for yoga and pilates studios: class schedule, bookings and package sales.
Plan · Claude

Problem

Bookings are scattered across WhatsApp and spreadsheets

Audience

Independent studios with 1–5 instructors

Value

A studio sets up its schedule in 10 minutes

  • MVPOne studio's class schedule and online booking6 steps
  • v1Package sales, student management, reminders6 steps
  • LaunchDomain, payments, legal pages, SEO6 steps

ST-4 · Agent prompt

Student booking page: 7-day schedule, waitlist for full classes, prevent overbooking with a transaction.

02 · Design

Your screens are designed in UIForge, and your agent codes from the design

UIForge is EnsaTech's design tool. Move your project there in one click: first a design language is built for your project, then every screen is dressed in it. The screens sit on your ShipTrack canvas as cards; click one for a live preview.

  • Your coding agent pulls the colors, Tailwind setup and screen code itself with get_design over MCP
  • Even a 30-screen app looks like it came from one hand: every screen uses the same token set
  • Design progress (how many screens are ready) shows on the project page and the canvas
StudioFlow · Design

Design: 3/4 screens ready

Open in UIForge

  • Home

  • Class schedule

  • Packages

  • Booking

get_design(screen: "class-schedule")

→ theme, tailwind config and 3 files handed to the agent

03 · Otopilot

Have the agent on your own computer do the steps

A small command-line tool runs on your computer and hands the step you send from the site to Claude Code, Codex, Cursor or Antigravity. The agent works with your account, in your folder.

  • Watch the agent live: which files it reads, what it edits, which tests it runs
  • Run all remaining steps of a version in one session; each finished step is ticked immediately
  • If it hits a usage limit, it picks up where it left off once the limit resets
Otopilot · LAPTOP-EV
ST-10…ST-123 steps in one session6 dk
  1. ST-10Student management and CRM
  2. ST-11Notifications and remindersnow
  3. ST-12Mobile-friendly student experience

Now: editing reminders.ts

  • Inspected 5 files
  • Edited students.ts and page.tsx
  • Ran the type check
  • ST-10 completed
  • Now adding the pre-class reminder emails.

04 · Keys

No more guessing which API key is missing

Otopilot scans your project folder and lists the environment variables your code needs. It shows where to get each missing one and writes the value you enter into your .env before the work starts.

  • Values are stored encrypted and never shown to your browser again
  • Direct “where to get it” links for services like Stripe, Resend and Supabase
  • If the agent gets stuck on a key, it tells you which one is missing
StudioFlow · Keys

2

Missing

1

Stored

1

On your computer

  • STRIPE_SECRET_KEY

    Stripe

    Stored …f456
  • RESEND_API_KEY

    Resend

    MissingWhere to get it
  • DATABASE_URL

    Database

    On your computer
  • AUTH_SECRET

    App

    Missing Generate

Values are encrypted; Otopilot writes them into your .env before each job.

05 · Revenue

After launch, same place: is money coming in?

Connect your Stripe (read-only key) or Lemon Squeezy account. Every day ShipTrack pulls monthly revenue, subscribers and failed payments, right next to your project.

  • Charts for monthly revenue, subscriber count and churn
  • Failed payments are listed separately, so you notice before you lose the customer
  • Your keys are read-only; ShipTrack can't take payments or move money
StudioFlow · Revenue

Monthly revenue

$1,276 +18%

Active subscribers

34 +5

Failed payments

2

  • Pro plan · monthly$39Paid
  • 10-class pack$85Paid
  • Pro plan · monthly$39Card declined

Figures shown are examples.

Agents

Works with the agent you already use

An agent connected over MCP reads your roadmap, ticks steps itself and pulls your UIForge design. The CLAUDE.md / AGENTS.md rule for your repo comes ready-made. Otopilot sends steps from the site in one click.

  • Claude Code

    Anthropic

    MCP + Otopilot

  • Codex

    OpenAI

    MCP + Otopilot

  • Antigravity

    Google

    MCP + Otopilot (full access)

  • Cursor

    Anysphere

    MCP

Security

You stay in control.

Trusting a tool that runs code on your computer isn't easy. That's why every default is the most cautious one.

Your code stays on your computer

The agent runs on your machine. ShipTrack only receives the step summary, the agent's output and the diff of the changes.

Every job starts with your approval

Otopilot asks on your computer before it starts a job. If you want it to start automatically, you turn that on.

Full access behind three locks

The mode where the agent runs commands without asking only works if it's enabled on the site, separately allowed on your computer, and confirmed every single time.

Encrypted keys

API keys are encrypted with AES-256-GCM. Connection tokens are only stored as hashes.

Row-level access

Database rules make sure every user only sees their own projects; no other account can reach your data.

GDPR and KVKK

No advertising or tracking cookies. Our privacy policy states plainly what data we process and why.

FAQ

Frequently asked questions

Can't find your answer? destek@ensatech.dev

Does ShipTrack write the code itself?+

No. The coding agent you use (Claude Code, Codex, Antigravity) writes the code on your computer. ShipTrack creates the plan, hands steps to the agent, tracks progress and shows you the result.

Where do I design how the app looks?+

In UIForge. The “Design with UIForge” button on the project page moves your project there; screens are built with a single design language and show up on your ShipTrack canvas. Your coding agent pulls the design itself with the get_design MCP tool, no copy-paste needed.

How does the agent know which step it's on and what it finished?+

The Connect page gives you a ready-made CLAUDE.md / AGENTS.md rule. With it the agent connects to ShipTrack over MCP, takes the next step and marks it done itself; the tick turns green on your board. Just tell your agent “do the next step”.

Do I need a Claude Code or Codex subscription?+

For Otopilot and the MCP connection, yes: agents run with your own account and usage counts against it. Creating the plan uses ShipTrack's own AI usage; you don't need anything else for that.

Does my code or any of my files go to your servers?+

No. Otopilot runs the agent on your computer and only sends us the step summary, the agent's output and the diff of the changes. Unless you enter them on the Keys page, environment variable values never leave your computer either.

Do I need to use a terminal?+

You run three commands once to set up Otopilot; the site gives them to you ready to copy, with your own key. After that, everything is managed from the site.

Does it cost anything?+

The free plan includes one active project at a time and three plans a day. A paid plan with more projects and plans is coming soon.

What happens if a step gets stuck halfway?+

If the agent can't continue because something is missing, it writes down why; if it's a missing API key, it says which one. Add the key and resume, or skip the step and carry on with the rest.

If your idea is ready, your plan is two minutes away.

Create an account, describe your idea, send the first step to your agent.

Start for free