Managing apps on the App Store involves a lot of repetitive work. You check review sentiment, update metadata, create in-app purchases, manage TestFlight testers, track build statuses — and every one of those tasks requires navigating through multiple screens, filling out forms, and clicking through confirmation dialogs. Even in a fast native app like Forge, some workflows still involve more steps than they should.
Forge Agent changes that. Instead of clicking through menus, you type what you want in plain English, and the agent does the rest.
What Is Forge Agent
Forge Agent is an AI assistant built directly into Forge. It lives inside the app as a chat interface — you type a request, and the agent reads your live App Store Connect data, performs actions, and reports back in real time.
This is not a chatbot bolted onto a dashboard. Forge Agent has deep access to the same API layer that powers every feature module in Forge. When you ask it to show your one-star reviews, it calls the same code that the Customer Reviews screen uses. When you ask it to create a new in-app purchase, it goes through the same pipeline that the IAP editor uses. The agent is not a separate system — it is another way to interact with the same tools you already have.
Forge Agent is powered by OpenAI and runs with your own API key. Your key is stored in your macOS Keychain, the same secure storage macOS uses for passwords and certificates. The agent communicates directly with OpenAI's API from your machine. There is no intermediary server, no Forge account, and no data passing through our infrastructure.
What Forge Agent Can Do
Forge Agent covers seven feature modules, each with a set of purpose-built tools that let it read, create, update, and manage your App Store Connect data.
Customer Reviews
Ask the agent to surface the reviews that matter. "Show me all one-star reviews from the last week" filters your reviews by rating and date. "Are there any reviews mentioning crashes?" searches review text for keywords. Once you see a review you want to address, you can ask the agent to draft and submit a response — or write your own and tell the agent to post it. The agent can also filter by territory, sort by date or rating, and paginate through large review sets.
App Info
Your app metadata — subtitle, description, categories, pricing, availability — all of it is editable through conversation. "Update my app subtitle to 'The fastest way to ship'" tells the agent exactly what to change. "Set my app price to tier 2 in the US store" adjusts pricing. The agent handles localization fields, content rights, age rating declarations, and app availability across territories. Instead of navigating through App Store Connect's nested forms, you describe what you want and confirm the change.
App Versions
Version management is one of those tasks that feels simple but involves a surprising number of status checks. "What is the status of my latest build?" gives you an instant answer. The agent can show you version history, release notes, build details, and submission statuses. When you need to track where things stand across multiple versions, a single question replaces several screens of navigation.
In-App Purchases
This is where Forge Agent saves the most time. Managing in-app purchases in App Store Connect is notoriously tedious — creating products, setting prices across regions, adding localizations, generating offer codes. The agent handles all of it.
"Create a new consumable IAP called 'Premium Credits' with a reference name of 'credits_100' and a price of $4.99" does exactly what it says. "Show me all my in-app purchases" lists your products with their types, statuses, and prices. "Add a promotional offer to my 'Pro Unlock' product" walks through offer creation. "Generate 500 offer codes for 'Premium Credits'" creates the codes you need for a marketing campaign.
The agent supports the full lifecycle: creating products, editing metadata, managing pricing and localizations, setting up promotional and subscription offers, and generating offer codes. It covers consumables, non-consumables, and non-renewing subscriptions.
Subscriptions
Subscription management follows the same conversational pattern. "Show me all active subscription groups" gives you an overview. "What subscriptions are in the 'Pro' group?" drills into a specific group. The agent can create and edit subscriptions, manage pricing across territories, configure introductory and promotional offers, handle subscription availability, and navigate the relationships between groups, subscriptions, and offers.
If you have ever tried to set up a subscription group with multiple tiers, regional pricing, and introductory offers through the App Store Connect web interface, you know how many clicks that takes. With Forge Agent, you describe the configuration you want, review the changes, and confirm.
TestFlight
TestFlight management through conversation is straightforward. "Add these testers to the beta group" takes a list of emails and adds them. "What is the latest build status?" checks processing. "Show me the beta feedback for build 247" pulls up tester feedback for a specific build.
The agent can manage builds, beta groups, individual testers, test information, and beta app review submissions. For teams that ship frequent betas, being able to manage testers and check build statuses without leaving your current workflow is a meaningful time saver.
Dashboard
The Dashboard module gives the agent a top-level view of your apps. "Show me all my apps" lists everything in your App Store Connect account with their current statuses. This is useful as a starting point — you can ask for an overview and then drill into any specific module from there. The agent knows which app you are currently viewing in Forge, so it defaults to your current context but can switch when you ask.
How It Works
Under the hood, Forge Agent is built on a tool-based architecture. The agent has access to 26 tool definitions spread across the seven modules. Each tool is a specific, well-defined action — things like "search reviews by rating," "create an in-app purchase," or "list TestFlight testers in a group."
When you send a message, the agent analyzes your request and decides which tools to call. It can chain multiple tools together in a single response — for example, first loading your subscription groups, then drilling into a specific group, then showing you the pricing for a particular subscription. The conversation flows naturally, and the agent maintains context between messages so you do not have to repeat yourself.
Responses stream in real time. You see the agent's thinking and results as they arrive, not after a long wait. Tool calls execute against your live App Store Connect data through Apple's official API, so what you see is always current.
The agent is also context-aware. If you are viewing the Customer Reviews module for your app "PhotoSync," the agent already knows which app you are working with. You do not need to specify "for PhotoSync" in every request — it picks up the context from your current Forge session. If you want to switch to a different app, just tell it.
Safety First
Forge Agent is designed with a strict safety model. Not all actions are created equal — reading reviews is harmless, but deleting an in-app purchase is not.
Read operations execute immediately. When you ask the agent to show you data, it fetches and displays it without any friction. There is nothing to confirm because nothing is being changed.
Write and create operations require confirmation. When the agent needs to modify or create something, it pauses and shows you exactly what it is about to do. You see the specific changes — "Update app subtitle from 'Old Subtitle' to 'New Subtitle'" — and explicitly approve or reject. Once you approve a write operation in a given turn, subsequent writes in that same turn auto-approve to avoid repetitive confirmation dialogs. But each new conversation turn resets the gate.
Delete operations always require confirmation. Every single time. There is no auto-approval for deletions, no matter what you approved earlier in the conversation. The agent will always stop and ask before removing anything.
Submission operations always require confirmation. Submitting a build for review or pushing a change live is gated independently from regular writes. Even if you approved edits earlier, the agent will pause again before any submission.
Your OpenAI API key stays in your macOS Keychain. Forge connects directly to Apple's App Store Connect API — the same connection it uses for all its features. There is no Forge server in the middle, no data routing through third-party infrastructure. The agent's actions are the same API calls you would make manually through the Forge interface, just assembled and executed through conversation instead of clicks.
Getting Started
Using Forge Agent takes about thirty seconds to set up:
- Open Forge and navigate to any feature module — Customer Reviews, In-App Purchases, Subscriptions, TestFlight, or any other
- Click the Agent chat icon in the toolbar
- Enter your OpenAI API key when prompted (stored securely in Keychain — you only do this once)
- Type your request in plain English
That is it. No configuration files, no prompt engineering, no learning a command syntax. If you can describe what you want to do, the agent can do it.
Start simple. Ask it to show your recent reviews, or list your in-app purchases, or check your latest build status. As you get comfortable, move to more complex requests — creating products with specific pricing, managing subscription offers, bulk-adding TestFlight testers. The agent handles single-step and multi-step workflows equally well.
Try It Today
Forge is free during launch, and Forge Agent is included. If you manage apps on the App Store, this is the fastest way to go from "I need to check something" to having the answer — or from "I need to change something" to having it done.
Download Forge and try Forge Agent for yourself. You can explore all of Forge's capabilities on the features page, or go straight to the download page to get started.
The next time you need to check your review sentiment, update a price, or add a tester to your beta group, just type it. The agent handles the rest.