Ploa in your agent

Ploa's MCP server gives your coding agent about 100 tools to take your app to the App Store and Google Play: set up the stores, build, sign, send to testers, write the store page, answer the privacy forms, and ask you before anything goes to App Review or production. You talk to the agent; the agent talks to Ploa.

What it is

MCP (Model Context Protocol) is the open standard coding agents use to call outside tools. Ploa ships an MCP server, so the agent you already work with (Claude Code, Cursor, VS Code with Copilot, Codex, Claude on the web, or any agent that speaks MCP) can work with your apps in Ploa the same way you do on the dashboard.

The agent doesn't get your store keys. It asks Ploa, Ploa calls Apple and Google from its server, and anything that can't be undone waits for your approval.

5 minutesConnect your agentOne command for agents on your computer, or a web address for agents on the web. RegistryEvery MCP toolEach tool by stage: what it does, when, whether it asks you, and what to ask. 7 briefsWhat to askStep-by-step briefs the agent follows, and requests that work well. Your callAccess and approvalsWhat an agent may do on its own, what waits for you, and autonomous agents.

Two ways to run it

The same tools come in two forms. Most people use the first.

	On your computer	On the web
For	Claude Code, Cursor, VS Code (Copilot), Codex, Windsurf, Claude Desktop and any local agent	Claude on the web and other agents that can't run a command on your computer
How it starts	The agent runs `ploa mcp` (or `npx -y @ploa/cli mcp`) in your project	The agent connects to `https://ploa.dev/mcp`
Sign-in	`ploa login` once, or a token in `PLOA_TOKEN`	You approve the agent on ploa.dev (OAuth), choosing how far it may go
Which app	The app in the folder it works in (from `ploa.json`)	Named by its ID; `ploa_apps` lists them
Tools	All of them	All that work from Ploa alone: no build on this machine, preflight, local files or simulator screenshots. Builds run on GitHub.
Access	Up to production, or autonomous	Read; drafts and builds; or drafts, builds and requests: it asks you for App Review, production, prices and testers, and you approve each. It never changes the stores directly and can't be autonomous.

How a request flows

You ask your agent
"Send the app to TestFlight and Google Play internal testing."
The agent calls Ploa's tools
It reads where the app stands (ploa_setup_guide, ploa_shipping_report), checks the project (ploa_preflight), fixes what the stores would reject, and starts the build (ploa_release_start).
Ploa does the store work
It signs the build with keys that never leave its server, uploads it, and follows Apple's processing and Google Play until testers have it.
You approve what can't be undone
App Review, production, prices, testers who get an email and replies under reviews wait for you on the dashboard or as a notification on your phone. The agent waits for the answer with ploa_approval_status.
Everything is recorded
Each step is in the release's run and in the workspace's audit log, under the agent's name.

Quick start

On the computer where you work, in the app's folder:

curl -fsSL https://ploa.dev/install.sh | sh
ploa login
ploa mcp install

Restart the agent, then ask it:

Ship this app to the App Store and Google Play with Ploa, and show me the steps that are mine.

The agent links the folder to the app in Ploa itself. Other agents, npx, settings by hand and the web server: Connect your agent. The whole path for a first app: Getting started.

What it can do

Every stage of an app's way to the stores, in the order it goes through them. Each card opens that stage in Every MCP tool.

Start and setupWhat's left in each store, who does each step, the app's ID and the Android upload key. Steps only you can doStore forms with no API, as tasks with numbered steps and links. Store detailsText in every language, category, age rating, price, App Review's contact. Privacy and export complianceApp Privacy, Data safety and Apple's encryption questions, worked out from the code. Screenshots and videosThe exact sizes, captured from a simulator, checked and uploaded in order. BuildsPreflight, builds here, on GitHub, in Xcode Cloud or your Google Cloud, and stopping one. TestersTestFlight groups and testers, What to Test, beta review, Play testing tracks. App Review and productionSubmit, withdraw, release, phased release and Play rollouts, each approved by you. RejectionsWhat the stores said, why, and the way back. Purchases and subscriptionsProducts, prices, free trials, offers, offer codes and RevenueCat. Reviews and salesCustomer reviews and replies, downloads, the store funnel and proceeds. Marketing and eventsCustom product pages, A/B tests and In-App Events. GitHubBuilds on GitHub Actions, PR labels, releases and issues. The webA site or web app at a free ploa.app address, tested and released like a build. The app's serverAn API, scheduled jobs and a database for the web app. Status and reportsWhere everything stands, each run step by step, and how stable a release is.

What the agent always does

Asks how it should work, first. An autonomous agent or approvals: your choice, asked once, and only you can turn autonomy on, in Ploa. Choosing at the start.
Starts from Ploa's guide. Every instruction for you (a key to make, a form in a store's console, a capability to turn on at Apple) comes from ploa_setup_guide. It says exactly what to do and matches the stores as they are today, with clickable links that open the exact page in App Store Connect or Play Console wherever Ploa knows the app's IDs there.
Asks, doesn't assume. The store page's languages, the countries, the price, who the app is for and the contacts are your decisions. The agent asks and waits.
Keeps you oriented. It shows the stages board at the start and after every step: what was done and by whom, what's left, and what's needed from you.
Says what Ploa can't do. Apple's App Privacy and most of Play Console's App content forms are entered by you in the console. The agent tells you exactly what to enter and where.
Waits for approvals. It never looks for another way around an approval.

Ploa has no AI of its own. The store text, screenshots and privacy answers are written by your agent, from your app's code. Ploa checks them against each store's rules and sends them once you say so (or an agent you made autonomous does).