Generate App Store Screenshots Automatically with an AI Coding Assistant: An Open-Source Tool Hands-On
Open-source skill lets AI coding assistants build multi-size, multilingual App Store screenshot editors locally.
The open-source project App Store Screenshots on Product Hunt hands App Store screenshot design to AI coding assistants like Claude Code, Codex, or Cursor. Users install the skill with a single npx command, provide screenshot assets and a short brief in their project, and the agent builds a local screenshot editing environment. It includes 18 styles, a Style Lab side-by-side comparison interface, copy editing, and export of full sizes for iPhone, iPad, Mac, and Android plus multilingual versions. It's completely free, open source, and runs locally—you only bear the agent's own invocation cost. The article also notes limitations: no actual output demos, design quality yet to be verified, and users needing basic command-line and AI coding tool experience.
App Store screenshots are a crucial part of an app's conversion rate, but designing a set of screenshots across multiple sizes and languages often requires a designer or a paid subscription tool. An open-source project called App Store Screenshots has appeared on Product Hunt, with the idea of handing screenshot design directly to your AI coding assistant—Claude Code, Codex, or Cursor.
What Problem It Solves
In the traditional workflow, developers hand app screenshots to design tools or contractors, then manually export each size variant for iPhone, iPad, Mac, and Android one by one. As language versions multiply, the workload grows exponentially. Paid screenshot tools charge monthly subscriptions, which is an ongoing cost for indie developers and small teams.
This project's approach is to have an AI coding assistant build a screenshot editor directly in your local project. You only need to provide your app screenshot assets and a short brief, and the agent will generate a usable editing interface. From there, style selection, copy editing, and multi-size export all happen within this local tool.
According to the Product Hunt page, the tool is completely free and open source, runs locally, incurs no additional subscription fees, and you only pay the invocation cost of the agent you use.
Core Features and Usage
Installation and Launch
Install with a single command:
npx skills add ParthJadhav/app-store-screenshots
After installation, launch Claude Code, Codex, or Cursor in your project, provide the screenshot files and a short requirements description, and the agent will build a screenshot editing environment within the project.
npx skills add is an extended use of the npx command in the npm ecosystem: npx temporarily downloads and executes an npm package—here the package is named skills, and its add subcommand takes a GitHub repository path (ParthJadhav/app-store-screenshots). Such commands typically pull the skill definition files from the repository (such as prompts, scripts, or workflow descriptions) into a conventional directory in the local project so the AI coding assistant can recognize and invoke them. Understanding this helps clarify the tool's actual form—it is not a standalone app, but a set of instructions and resources for the agent to read, so it must be run within a specific project context to take effect.
Styles and Comparison
The tool includes 18 built-in styles covering a range of visual designs. It also provides a comparison interface called Style Lab, where you can view different styles side by side to quickly pick the one that fits your app's personality.
AI coding assistant (Agent) here refers to tools like Claude Code, Codex, and Cursor that have file read/write, command execution, and multi-step task planning capabilities. What distinguishes them from ordinary code completion is their ability to autonomously complete the loop of "read requirements → generate code → run and verify → iterate." This project leverages exactly that: the agent generates an interactive screenshot editor frontend locally based on the brief, rather than a server returning fixed templates. This also explains why the cost is described as "agent invocation cost"—each generation or modification consumes token quota on the corresponding platform, and the expense depends on task complexity and the chosen model.
Copy Editing and Multi-Platform Export
After selecting a style, you can edit the copy on the screenshots and export all sizes for iPhone, iPad, Mac, and Android, with support for multiple language versions. For apps that need to be listed on multiple platforms or target overseas markets, this step saves a great deal of repetitive work.
The practical significance of multilingual screenshot export lies in the localized display within app stores: the App Store and Google Play show different screenshot assets based on user language and region. If you only provide a single-language version, conversion rates in non-native markets will drop noticeably. The traditional approach is to lay out and export a full set of sizes separately for each language, whereas this tool separates the copy layer from the style layer, so switching languages only requires replacing text and reusing the layout for batch export. Note that each platform has its own specifications for screenshot dimensions, aspect ratios, and counts (e.g., iPhone and iPad have different aspect ratios). The tool claims to cover all sizes, but it's still advisable to verify against official documentation after export to confirm compliance with the latest store requirements.
Why It's Worth Noticing
The value of this project isn't just a set of screenshot templates—it demonstrates a "skill-ified" tool form: packaging the workflow of a specific vertical scenario into a capability that an AI agent can invoke, with the agent generating a dedicated interface locally. Compared to fixed SaaS products, this model is more flexible and fits better into developers' existing work environments.
From a cost perspective, the combination of open source and local execution eliminates the long-term expense of subscriptions, which is especially friendly to budget-conscious indie developers. Its 76 upvotes and #12 spot on the Product Hunt homepage show that this kind of need has real resonance in the developer community.
Things to Keep in Mind
The project description doesn't include actual screenshot demos of its output, nor does it explain the specific design quality of the 18 styles. For a design-oriented tool, whether the final output meets store-ready standards can only be judged after hands-on trial. In addition, usage depends on the AI coding assistant you already have, and agent invocation costs vary by platform and usage frequency—the description lists it as "the usual agent cost."
There's also an implicit barrier: users need basic command-line skills and experience with AI coding tools. Developers who are completely unfamiliar with Claude Code or Cursor will need to get comfortable with those environments first.
Who It's For
Indie developers, small teams, and teams that frequently update their app versions could all benefit from this tool. If you already use Claude Code, Codex, or Cursor in your daily development, adding it to your workflow as an invocable skill involves almost no extra learning cost.
The project author is Parth Jadhav, and the tool is released as open source. The project entry point and installation command can be found on the Product Hunt page.
Related articles
Microsoft Fabric Database Agent in Action: From Reactive Firefighting to Intelligent Detective
A deep dive into the Microsoft Fabric Database Agent: how to enable it, supported monitoring scenarios, a blocking-issue demo, and the roadmap from reactive to autonomous operations.
Zendesk Uses ChatGPT Enterprise to Build a Sales Agent with 1,000 Daily Active Users
Zendesk opened ChatGPT Enterprise to all employees, who built about 600 apps themselves, including a sales expansion agent with about 1,000 daily active users. This article analyzes how it turns internal data into sales opportunities and its "don't wait" implementation strategy.
Google Releases Gemini 4 Carbon: Its Strongest Model Yet, Debuting with a Cybersecurity Focus
Google releases Gemini 4 Carbon, calling it its strongest model, with a cybersecurity focus. It tops the third-party composite ranking at 59 points but leads by only one or two. This article analyzes its security test results, staged rollout, token pricing, and the divergence between self-reported and third-party tests on hallucinations.