42 free agent skills take an app from code to the stores
Building an app with AI is easy now; getting it signed, listed and through review still stops people. App Factory packs the whole last mile into 42 open-source agent skills that load only where you need them.
Published Oct 23, 2026
The Wow
What someone built
Developer yybd's App Factory (MIT) is a Claude Code plugin marketplace with tracks for the App Store (16 skills), Google Play (9), shared work such as naming, copy and release order (7), web, Capacitor, design, Windows and launch posts. Orchestrator skills sequence a full release; executors build, sign, upload and write icons in every size.
- Made by
- yybd on GitHub (not us)
- Original repository
- App Factory on GitHub
- License
- MIT License
- What you need
- Claude Code
- Python 3
- Xcode on macOS for Apple tracks
- A JDK for Android tracks
- fastlane for store uploads
- Install
- In Claude Code run `claude plugin marketplace add yybd/app-factory`, then in your app's folder run `python3 ~/.claude/plugins/marketplaces/app-factory/factory/enable.py` and the install lines it prints. QUICKSTART.md in the repo walks one app through in about fifteen minutes.
- A business you could build on it
- Store release desk for AI-built apps
Why it's remarkable
It covers the boring steps everyone forgets, in the right order, with firm limits: it reads credentials in place, never submits for review, never changes a price and never pushes.
The Opportunity
The business it reveals
A fixed-price release desk for founders with a working app and no store listing: listings, screenshots, signing and uploads on the client's own accounts, then a smaller fee for each update.
Who would pay
Non-technical founders with AI-built apps, small studios shipping client apps, and indie developers selling desktop tools outside the stores.
The Flip
3 ways you could turn this into a business
Store release desk for AI-built apps
- Who pays
- Non-technical founders and small businesses with a finished app
- How it would earn
- A fixed fee per first launch and a smaller fee per update (our hypothesis; price untested)
- Why it might work
- Building got easy; releasing didn't. The skills cover the steps founders get stuck on, and the client keeps the accounts.
- What already exists
- Freelance app-release gigs on Upwork and Fiverr, mobile CI services such as Codemagic and Bitrise, and fastlane for developers who set it up themselves.
Release operations for small app studios
- Who pays
- Agencies and studios shipping apps for several clients
- How it would earn
- A monthly retainer per client app (our hypothesis)
- Why it might work
- Studios repeat the same release chores for every client; per-repo tracks keep each app's rules separate.
- What already exists
- An in-house release engineer, CI pipelines and fastlane scripts maintained by the studio.
Direct-download setup for desktop indie tools
- Who pays
- Indie developers selling Mac or Windows apps outside the stores
- How it would earn
- A one-off setup of signed installers and self-updates (our hypothesis)
- Why it might work
- The Apple and Windows tracks cover distribution outside the stores, a chore many indie developers postpone.
- What already exists
- Electron and Tauri updater docs, commercial installer tools, and freelance desktop developers.
Why we'd try "Store release desk for AI-built apps" first: There's a fast-growing crowd of people with working apps and no listing, and this toolkit covers most of the steps between them and the store.
Build It
How you could build it
A release desk: intake form, account-access checklist, listing and screenshot pack, signed builds uploaded to the client's accounts, and a handover checklist.
First step: Follow QUICKSTART.md with an app of your own (or a sample app) and time a full TestFlight or internal-testing release.
Agents and tools
- Claude Code
- App Factory
- Xcode, a JDK and fastlane
- The client's store accounts
Setup
- Enable only the tracks a client's repo needs.
- Run a full test release on your own app and note every manual step.
- Write the client access checklist (team roles, never shared passwords).
- Price a first launch and an update.
- Offer it in two AI-builder communities you already use.
What the agent does
- Checks names against the store
- Writes icons in every size and listing copy
- Builds, signs and uploads
- Sequences the release and flags missed steps
What you do
- Gets client access the safe way
- Reviews every listing and screenshot
- Submits for review and handles rejections
- Owns the client relationship
Your first customer: A founder who has posted a working app demo but has no store link.
Test demand first: Offer five founders a fixed-price first launch. Success: two say yes and one release reaches internal testing within a week. Failure: nobody wants to hand over access, which says the offer needs a different trust step.
Your starter instructions
Paste this into the AI agent you use (ChatGPT, Claude, Gemini, Grok Bot or another) as your first message. Copying it starts work in your own agent; it doesn't create or deploy a business by itself.
Running it
What stays human: Client onboarding, store-account access (the client owns the accounts), reviewing listings and screenshots, submitting for review, answering reviewer questions and fixing rejections.
- Startup cost (estimate)
- About $0 beyond a Claude Code plan and a Mac (estimate) (Estimate: App Factory is MIT-licensed and fastlane is free; Apple builds need Xcode on macOS; store developer fees are paid by the client on their own accounts. Not measured)
- Running cost
- Agent usage per release (estimate pending a pilot) (Unknown until a pilot times a few releases. Not measured)
- How much the agent runs
- People run it; AI helps with tasks.
Pricing your own version? Use the free startup-cost worksheet.
The evidence
What the source shows: The repository's README, TRUST and QUICKSTART notes (created Sep 29, 2026; updated Oct 23). We read them; we haven't run a release.
What's ours: The release desk, pricing and flips are our application.
Worth planning around
- Built for and tested only in Claude Code.
- Store review can still reject an app for reasons the skills can't fix.
- Client accounts and signing keys must stay with the client.
- The project is young and changing quickly.
Sources
Sources reopened Oct 23, 2026. Spot a mistake? Tell us and we'll correct it in public.
Keep reading
- Services to sell to people building apps with AI (guide)
- How much of the work do AI agents actually run? (guide)
- He still can't write code. His five-person ad agency now runs on about 270 tools he built with Claude. Every owner-run firm needs its first five
- A marketer who doesn't code built an elk-call coaching app with Claude Code and says it hit a $650K run rate in 110 days. Every skill learned by sound is the same opening
- A free n8n workflow flags fake supplier invoices before anyone pays them — and Germany's e-invoice switch makes it a service bookkeepers can sell
- Marc Lou's agent chased a broken-shoe refund through three support desks and got his 850 AED back. Small stores now need a claims desk that can answer agents
- Every discovery and launch plan