Store screenshots, from your coding agent.
Ask Claude Code, Cursor or any MCP client for App Store or Google Play screenshots. It sends your app's real screens, and you get back a link to three finished directions — fully editable, in any store language.
No key needed to try it — your first set is free.
claude mcp add --transport http launchshots https://launchshots.com/api/mcpThen type /mcp__launchshots__store_screenshots, or just ask for store screenshots.
With an API key the same line carries an Authorization header — account settings gives you the finished version when you create one.
- Step 1
Your agent gathers the screens
It reads your README and code to understand the app, then picks your key screens — from existing screenshots or captured fresh from the simulator.
- Step 2
LaunchShots designs the set
The same designer as the homepage reads every screen, orders them, writes a headline for each and art-directs three ways. About a minute or two.
- Step 3
You pick a direction
Your agent hands you a link. Choose a direction and it opens in the editor, where every headline, colour and frame is still editable and exports at every store size.
- “Make App Store screenshots for this app.”
- “Create Play Store screenshots in German — lead with the offline mode.”
- “Take screenshots of the onboarding, dashboard and settings screens in the simulator, then turn them into a store set.”
- “Generate the iPhone and iPad sets, dark and premium.”
In Claude Code the whole flow is also a slash command: /mcp__launchshots__store_screenshots
| No key | API key | |
|---|---|---|
| Sets | A few a day per connection, from a shared allowance | 3 free, then your credits or plan |
| Keeping it | Free signup from the review link | Opens straight into your projects |
| Tablet set | — | Yes, as a second generation |
| Who can open it | Whoever claims the link first | Only your account |
Endpoint https://launchshots.com/api/mcp — Streamable HTTP, stateless. Authenticate with Authorization: Bearer ls_key_…, or leave it off for the free tier.
prepare_screenshot_uploadStarts a set and returns one upload URL per screenshot, with ready-to-run curl commands.
generate_screenshot_setChecks the screenshots, takes the generation from your allowance and starts designing. Returns the review link straight away. Screenshots can also be sent inline for agents without a shell.
get_screenshot_setWaits for the set and returns the review link, the headline written for every screen and the three directions.
get_accountWhich account the connection uses, and how many generations are left.
list_projectsRecent projects on the account, with editor links. Needs an API key.
store_screenshotspromptThe whole workflow as a reusable prompt — understand the app, gather the right screens, generate, and hand back the link.
- Do I need an account?
- Not to try it. Without an API key your agent can make a set drawn from a small shared daily allowance, and you create a free account from the review link to keep it. With a key, sets are tied to your account and use its generations.
- What does it cost?
- The same as generating in the app. Every account gets 3 free generations; after that each one is 1 credit, or included (up to 50 a month) on the Launch Kit and unlimited plans. A run that fails is refunded automatically. The tablet set counts as a second generation.
- Which screenshots should my agent send?
- Three to eight of the screens that matter most — one per core feature — with realistic content in them. Empty states, login walls and debug banners make weak headlines. The server tells agents this, but it helps to say which screens you care about.
- What happens to my screenshots?
- They are stored privately and used only to design your set. A set nobody opens is deleted after 14 days (30 with an API key). When you open a set, the screenshots are copied into your project like any upload.
- Can someone else open my review link?
- Anyone with the link can see the set, so share it like you would a private doc link. A set made with an API key can only be opened into the account that key belongs to.
- Something went wrong — was I charged?
- No. Any run that fails, times out or is refused is handed back, and your agent can retry the same set.