# Connect your AI to ASO Studio

Source: https://asostudio.app/agents

ASO Studio ships an MCP server, so an AI agent can read your keywords, rankings and market, research live App Store data, and make the changes you approve.

# Connecting an AI to ASO Studio

ASO Studio is a native macOS app for App Store keyword research. It runs an MCP
server, so an AI client can read the user's real keywords, rankings, market and
release plans, research live App Store data, write the listing and ship it to App
Store Connect, and run Apple Search Ads.

There are two ways to connect, and they differ in one respect only.

- **Anywhere**: the client talks to `https://asostudio.app/mcp`, and the user's Mac holds an outbound
  connection to that address. Needs an ASO Studio subscription, a sign-in with
  Apple, and the Mac awake.
- **Only on your Mac**: the client launches a program that talks to the running app.
  For clients on that machine only, not over the web. No sign-in.

Either way the work happens in ASO Studio on the user's Mac, so App Store requests
come from their machine and the workspace never leaves it.

## Anywhere

### Claude Code
In the terminal, from any folder.

1. Run this once.

   ```
   claude mcp add --transport http aso-studio https://asostudio.app/mcp
   ```
2. Start Claude Code, run /mcp, and choose to authenticate.
3. Sign in with Apple and approve the connection in the browser it opens.

Working when: Run /mcp. ASO Studio should read as connected rather than needing authentication.

### Claude Web
claude.ai in a browser. The desktop app takes the same address, in the same place.

1. Open Claude, then Settings, then Connectors.
2. Choose Add custom connector and give it this address.

   ```
   https://asostudio.app/mcp
   ```
3. Claude opens ASO Studio in your browser. Sign in with Apple and approve the connection.

Working when: The connector shows as connected. Ask it to list your apps and it should name them.

### ChatGPT Web
chatgpt.com with developer mode on. Plus, Pro, Business, Enterprise or Edu.

1. In ChatGPT on the web, open Settings, then Apps, then Advanced settings, and switch on Developer mode.
2. Back in Apps, choose Add custom connector and give it this address.

   ```
   https://asostudio.app/mcp
   ```
3. Approve the connection when ChatGPT sends you to ASO Studio to sign in with Apple.

Working when: The connector appears in Apps with its tools listed, and you can pick it in a conversation.

### Codex
A url rather than a command, which is what puts it on the remote transport.

1. Open ~/.codex/config.toml, creating it if it is missing.
2. Paste this at the end.

   ```
   [mcp_servers.aso-studio]
   url = "https://asostudio.app/mcp"
   ```
3. Save, then run this and approve the connection in the browser.

   ```
   codex mcp login aso-studio
   ```

Working when: Run codex mcp list. ASO Studio should be listed and authenticated.

### Your editor
Cursor, and any editor that takes a remote MCP server as JSON.

1. Open your editor's MCP settings and add a new server.
2. Paste this into the config file it opens.

   ```
   {
     "mcpServers": {
       "aso-studio": {
         "url": "https://asostudio.app/mcp"
       }
     }
   }
   ```
3. Save. When it first connects it will send you here to sign in and approve.

Working when: The server appears in your editor's MCP panel with its tools listed.

### Something else
Anything else that speaks MCP over HTTP.

1. Point it at this address as a remote MCP server.

   ```
   https://asostudio.app/mcp
   ```
2. Do not configure a token. The endpoint answers an unauthenticated request with where to sign in, so a client that follows the standard discovers the rest on its own: it registers itself, sends you here to approve it, and stores what it gets back.

Working when: Your client should offer to sign in rather than reporting an authentication error.

## Only on your Mac

### Claude Code
In the terminal, from any folder.

1. Run this once.

   ```
   claude mcp add --scope user aso-studio -- "/Applications/ASO Studio.app/Contents/MacOS/aso-mcp"
   ```
2. Start Claude Code in your app's project folder.

Working when: Run claude mcp list. ASO Studio should show as connected.

### Claude Desktop
The Mac app, talking to the Mac app.

1. In Claude Desktop, open Settings, then Developer, then Edit Config.
2. Paste this into the file that opens.

   ```
   {
     "mcpServers": {
       "aso-studio": {
         "command": "/Applications/ASO Studio.app/Contents/MacOS/aso-mcp"
       }
     }
   }
   ```
3. Save and restart Claude Desktop.

Working when: The tools icon in the chat box lists the ASO Studio tools.

### Codex
A command rather than a url, which is what keeps it on this machine.

1. Open ~/.codex/config.toml, creating it if it is missing.
2. Paste this at the end.

   ```
   [mcp_servers.aso-studio]
   command = "/Applications/ASO Studio.app/Contents/MacOS/aso-mcp"
   startup_timeout_sec = 60
   tool_timeout_sec = 300
   ```
3. Save, then start Codex.

Working when: Run codex mcp list. ASO Studio should be listed.

### Your editor
Cursor, and any editor that takes a local MCP server as JSON.

1. Open your editor's MCP settings and choose to add a new global server.
2. Paste this into the config file it opens.

   ```
   {
     "mcpServers": {
       "aso-studio": {
         "command": "/Applications/ASO Studio.app/Contents/MacOS/aso-mcp"
       }
     }
   }
   ```
3. Save. Most editors pick it up without a restart.

Working when: The server appears in your editor's MCP panel with a green dot.

### Something else
Anything else that launches a local MCP server.

1. Point it at this program as a local, stdio MCP server.

   ```
   /Applications/ASO Studio.app/Contents/MacOS/aso-mcp
   ```
2. No arguments, no environment, and no sign-in. It talks to the copy of ASO Studio already running.

Working when: Your client should list the ASO Studio tools once it has started the program.

## What it can do

- Write the title, subtitle and keyword field, and ship them to App Store Connect
- Create and update the listing in other languages
- Run Apple Search Ads: new campaigns (created paused), keywords, bids and pauses
- Plan the next release, ship it, and see how it did thirty days later
- Read sales, keywords, rankings and history
- Research the live App Store: which apps do the same job, and what their users ask for
- Track and untrack keywords and apps, tag them, write notes
- Answer the judgements ASO Studio would otherwise need an AI key for

## What it will not do

- Use a tool the user switched off when connecting it
- Spend money without showing the amount
- Write to the App Store without showing every field, before and after
- Work while ASO Studio is closed, because the work happens there

Every tool starts switched on and the user can turn any off. Whatever is on, a
change to the App Store listing or a deletion asks the user every time: in the
client if it can ask, in the ASO Studio window if not.

Full walkthrough: https://asostudio.app/agents


## Research your app with Claude, ChatGPT, or another MCP client

Connect a compatible AI assistant to ASO Studio and ask it to investigate your App Store market using the app's research. It can inspect competitors, read reviews, explore and measure search terms, and read your saved release plans. Your Mac must be awake with ASO Studio running, and the connection needs permission for the actions you request.

### An example research request

> Research my app in the US App Store. Find competitors that solve the same job, read recent reviews for recurring requests and complaints, and compare relevant search terms. Separate observations from estimates, name the source apps and storefront, and suggest one next step. Read only for now; do not change my listing or campaigns.

### Name the app and market
Give the assistant the App Store link or ID and the storefront. Ask it to confirm which app and workspace it is using before starting the research.

### Ask for evidence, then interpretation
Have it read competitors, recent reviews, and measured terms. Ask for source apps, original review wording, measurement dates where available, and gaps in coverage. The assistant's explanation should be distinguishable from the data ASO Studio returned.

### Review a specific next step
Ask it to explain a product opportunity or listing change using the findings. It can read saved plans and, with the relevant permissions and App Store Connect setup, help apply supported metadata changes after a preview. Review the actual diff before authorizing a write.

The result should connect a recommendation to the observations behind it: which apps were compared, what users asked for, which searches were measured, and what remains uncertain. This is an example request, not a precomputed report or a guarantee that every app has enough public evidence.

The research workspace lives on your Mac. Information returned to a connected assistant is shared with that assistant, so its own data handling also applies. A local MCP connection describes the transport; it does not mean a hosted model runs locally.

## Related research

- [App Store market research](https://asostudio.app/blog/app-store-market-research)
- [App review mining](https://asostudio.app/blog/app-review-mining)
- [App Store release planning](https://asostudio.app/blog/app-store-release-planning)
- [Trending App Store searches](https://asostudio.app/blog/app-store-trending-searches)
- [Apple Search Ads and organic rankings](https://asostudio.app/apple-search-ads)