StoreboardStoreboard/Agent access (MCP)
Let your coding agent do the ASO
Storeboard for Mac speaks MCP, the Model Context Protocol. Turn it on and Claude Code, Cursor, Codex — or any MCP client — can work in Storeboard the way you do: research keywords, write the listing and every screenshot caption in every language, drop in fresh simulator captures, and render the result to check its own work. It all happens in the project you have open, on your screen, and every change undoes with ⌘Z.
Off until you turn it on. Only this Mac can connect. Agents can't publish to App Store Connect or spend your AI key — those stay yours.
Works with
| Agent | How it connects |
|---|---|
| Claude Code | HTTP — one command in a terminal |
| Cursor | HTTP — an entry in ~/.cursor/mcp.json |
| Codex | stdio — through npx mcp-remote in ~/.codex/config.toml |
| Any MCP client | HTTP transport with a URL and header, or the same stdio bridge |
Storeboard writes the exact line for your agent — with your address and token filled in — in Settings → Agent access. For Claude Code it looks like this:
claude mcp add --transport http storeboard http://127.0.0.1:4977/mcp \
--header "Authorization: Bearer <your token from Settings>"Setting it up
- In Storeboard, open Settings → Agent access and switch on Let coding agents drive Storeboard (MCP).
- Pick your agent and copy the line Storeboard shows.
- Run it (or paste it into your agent's config), then open the Storeboard project you want the agent to work on.
- Ask. "What can I ask my agent?" in Settings has examples to start from.
What your agent can do
| Area | Tools |
|---|---|
| Projects | List and open projects, read a project's overview, import a .asoproj file |
| Listing | Read and write every field in any language — name, subtitle, promo, keywords, description, What's New; add languages, switch them on or off for export |
| Keywords | Live popularity, difficulty and opportunity for any storefront; Apple's own autocomplete ideas; track and archive terms |
| Competitors | Find apps, watch or unwatch them, read the daily change feed |
| Screenshots | Captions and subtitles per language; replace any screenshot, shared or for one language; apply a design template; render any screen exactly as it will export |
| Layout & brand 1.1.6 | Everything in the inspector: pan and zoom inside each phone, size, position, tilt, backgrounds, frames, caption styling, stickers and text, and the brand kit |
Twenty-four tools in 1.1.5. Version 1.1.6 adds six more, the second phone on a two-phone screen, and every device you design for — iPhone, iPad, Mac and Watch — without switching what you're looking at. A request an agent gets wrong is refused with the reason, never half-applied: text over Apple's limit, or a typo in a setting name, comes back as an error rather than a silent no-op.
Things to ask
- "Research keywords for my app in Germany and track the eight with the best opportunity."
- "Write Japanese captions for every screenshot, render each one, and shorten any that wrap badly."
- "Here's a new simulator capture — make it screenshot 3 in every language."
- "Nudge the front phone on screenshot 5 up so the status bar is hidden." 1.1.6
- "Which of my competitors changed their screenshots this week, and what did they change?"
How safe is it?
- Off by default. Nothing listens until you switch it on, and switching it off stops it.
- Only this Mac. The server listens on
127.0.0.1alone — no other computer, on your network or anywhere else, can reach it. - A private token. Every request must carry the token from your Settings; anything without it is refused. Regenerate the token to cut off every agent you connected before.
- In the open. Agents work in the project on your screen — each change appears as it happens, with a note saying what the agent did, is autosaved like your own edits, and undoes with ⌘Z.
- Things stay yours. Agents can't publish to App Store Connect, can't see your App Store Connect key, and can't run Storeboard's AI or spend your AI key. Publishing is a button you press.
- No new server. Storeboard still sends nothing to us. The one thing to know: your agent shares what it reads with its own AI provider (Anthropic, OpenAI, Cursor…) under that provider's terms — the same as any code it reads.
Questions
Does it cost extra?
No. Agent access comes with Storeboard for Mac and doesn't need Storeboard Pro. Your agent's usage is billed by its provider, as always.
Which version do I need?
Agent access is new in Storeboard 1.1.5. The layout and brand-kit tools arrive in 1.1.6.
Can an agent on another computer use my Storeboard?
No — only agents running on the same Mac as Storeboard can connect. Run Storeboard on the computer where your agent runs.
Is it on the iPhone app too?
No. Agent access is a Storeboard for Mac feature; coding agents run on the Mac.