Skip to main content
AppScreenshotStudio
MCP server + REST API

App Store screenshots,
from your agent or your terminal

Install the MCP server and Claude Code, Cursor, or Windsurf generate them as a native tool. Or call the REST API with three curl commands. Describe, refine, render: no browser, no design tools.

Get your API keyMCP setup guide
agent session
›
Works withClaude CodeCursorWindsurfCodex CLIZedcurlGitHub Actions+ any MCP client
Two ways in

Let your agent do it, or script it yourself

Same engine underneath. Pick the entry point that fits how you work.

For AI agents

MCP server

One install gives Claude Code, Cursor, Windsurf, Codex, and any MCP client 9 native tools: research your codebase, generate, iterate, upload your app screens, and render.

Claude Code plugin · easiest
/plugin marketplace add TWWorks-org/mcp-server
/plugin install appscreenshotstudio@appscreenshotstudio
Or add the MCP server
claude mcp add appscreenshotstudio -- npx -y @appscreenshotstudio/mcp
Or skill only · no MCP
curl --create-dirs \ -o ~/.claude/skills/appscreenshotstudio/SKILL.md \ https://appscreenshotstudio.com/openclaw/SKILL.md

Then just ask: “Generate App Store screenshots for my app.”

MCP setup guideBrowse on ClawHub
For scripts & CI

REST API

Three calls from any HTTP client: create a project, describe the screenshots, render to PNGs. Wire it into GitHub Actions, Fastlane, or a shell script.

terminal
API=https://appscreenshotstudio.com/api/v1
curl -X POST $API/projects -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" -d '{"device_id":"iphone-6.9"}'

Bearer-token auth. Rate-limit headers on every response.

REST API docs
9

MCP tools

3

REST calls, end to end

Free

Every render

6

iPhone, iPad, Android, Watch devices

The loop

Screenshots are a conversation, not a one-shot

You don't need the perfect layout up front. Describe what your app does, look at the options, and iterate until it ships. The API runs the whole loop.

01

Describe

Tell your agent what the app does. It reads your codebase for brand colors, screens, and copy.

02

See options

Get a full set back: layouts, headlines, device angles, trust badges. Look at what landed.

03

Refine

Point at what works. "Bolder headline, gold frame, tilt left." One message per pass.

04

Ship

Render to pixel-perfect PNGs at exact App Store dimensions. Free, as many times as you want.

One message each

Localize, switch device, or restyle in a sentence

Every card stays editable by chat. Ask your agent (or POST to the chat endpoint) and it happens across the whole set, not one card at a time.

Localize
› Translate all 5 screenshots into German, then make a Japanese set.
Switch device
› Duplicate this project for iPad Pro 13-inch. Rescale everything.
AI panoramic
› Pano an AI mountain-sunrise background across cards 1 to 3.
Use your screens
› Drop my Simulator screenshots into the device frames.
Frame color
› Make the iPhone frame gold across every card.
Restyle text
› Bold the headline and put "Free" in a green highlight pill.

Localization renders Latin, Cyrillic, and CJK scripts. Device clones rescale every element to the new canvas: iPhone, iPad Pro 13-inch, Android, or Apple Watch Ultra 2.

Copy, paste, done

The whole workflow, three ways

Whether an agent calls the tools or you script the endpoints, it is the same three steps: create, design, render.

# After installing the MCP server, in Claude Code / Cursor / Windsurf:

you    Generate 5 App Store screenshots for my running app.
       Brand color #10b981, iPhone.

agent  → generate-screenshots        (5 credits)
       5 cards ready. Want to iterate?

you    Make headline 1 bolder, tilt the phones left.

agent  → edit-screenshots            (5 credits)
       → render-screenshots          (free)
       5 PNGs at 1260 × 2736. Download ready.
Questions

Everything else

Yes. Install the MCP server and Claude Code, Cursor, Windsurf, Codex, or any MCP client gets 9 native tools. Ask "generate App Store screenshots for my app" and the agent researches your codebase, designs a set, iterates, and renders the PNGs for you.

For the agent path: anything that speaks the Model Context Protocol, including Claude Code, Cursor, Windsurf, Codex CLI, and Zed. For the REST path: anything that can make an HTTP request, including curl, GitHub Actions, Fastlane, and Jenkins.

No. The full loop (create a project, design with natural language, iterate, upload your own app screens, and render to PNGs) runs from your terminal, CI pipeline, or AI agent. The web app is optional.

Each design message costs 5 credits. Rendering is always free: render as many times as you want. API and MCP access come with the Growth plan ($19.99/mo, 200 credits) or Pro plan ($39.99/mo, 600 credits).

Yes. API keys need a Growth or Pro subscription. The Starter plan does not include API or MCP access. This keeps the service fast and reliable for everyone.

iPhone 16 Pro Max (1260 x 2736), iPhone 17 Pro (1206 x 2622), iPad Pro 13-inch (2064 x 2752), Android Phone (1080 x 2340), Android Tablet 7-inch (1200 x 1920), and Apple Watch Ultra 2 (410 x 502). Pass the device_id when you create a project.

Design (chat) is 60 requests per 5 minutes per key. All other endpoints are 120 requests per minute. Rate-limit headers are included in every response.

Ship from where you code

Screenshots shouldn't require a browser

Grab an API key, install the MCP server, and let your agent handle the App Store.

Get your API keyRead the MCP guide