Turn a Screenshot or Design Reference into a Working Mobile App

Role: you are a specialist in no-code software stacks.

Goal: within 7 days, build [type of software] with no-code tools only.

Lay out the full stack:
1. The specific tool that handles each component.
2. Why this combination beats the alternatives for my use case.
3. The build sequence, step by step.
4. The realistic feature set ready at launch on day 7 that people would genuinely pay for.

Role: you are a senior mobile developer who specializes in AI conversational interfaces, with support for several models and an enterprise-grade chat architecture.

Task: create a complete app for chatting with AI. It needs conversation management, support for multiple models and a refined dark UI that looks and feels native on any device.

Steps:
1. Before starting, ask me for my preferred AI provider, the endpoint configuration, the app name and any specific features I want.
2. Build the main chat screen: message bubbles, an input field, an indicator of the active model and the conversation history.
3. Create a model configuration screen where users switch between providers and adjust the parameters.
4. Build conversation management so users can create, name, organize and delete chat sessions.
5. Add streaming so the AI's reply appears one word at a time instead of all at once.
6. Add a system prompt builder that lets users create custom AI personas for different uses.

Rules:
- Dark interface everywhere; no light mode option unless I explicitly ask for it after the first build.
- Message history persists between sessions without user login and without cloud storage.
- On a slow connection, streaming degrades gracefully, with no lost message content and no corrupted history.
- Every API error is shown to the user as a helpful message, never as a technical error code.

Deliver the build in stages, in this order:
chat interface → model settings → management of conversations → streaming engine → system prompt builder → the full app

Role: you are a specialist in AI-powered browser tools.

Goal: I want to bring AI capabilities to [specific workflow] through a tool that runs in the browser.

Design the tool in four sections:
1. AI function: the specific thing the AI does that makes the tool feel truly magical for this use case.
2. Prompt architecture: how the prompts behind the tool's core feature are structured.
3. Interface: a user interface that makes the AI approachable for people with no technical background.
4. Pricing: a price that makes my tool feel worth paying for compared with simply using ChatGPT directly.

Example: Create a clean, nicely designed landing page for [agency name], a cold email agency artifacts

Template: [your request] artifacts

1. Phase 1 — Scoped build planCopy

   I want to create a simple app starting from nothing.
   
   - App idea: [app idea]
   - Who will use it: [target user]
   - What it should help them achieve: [what the app should help them do]
   - Core features: [3 to 5 simple features]
   - Preferred tech: [React / Next.js / plain HTML, CSS and JS / your choice]
   
   Before you write any code:
   1. Ask me any questions that matter for the build.
   2. Then draw up a simple build plan that covers the pages, the components, the data structure and what version 1 includes.
   
   Version 1 must stay small. I want an app that works before we add any advanced feature.

2. Phase 2 — Version 1 onlyCopy

   Following the plan we agreed on, build version 1 and nothing more: the smallest version of the app that works.
   
   Requirements:
   - the app loads
   - the main feature works
   - the design is clean enough to test with
   
   Leave out login, payments, a database, notifications, admin dashboards and every other advanced feature, unless one of them is truly essential.
   
   When the build is done, run the project, look for errors and fix whatever is broken.

3. Phase 3 — Fix the errorsCopy

   Run the app and examine the error.
   
   1. Explain in plain English what is broken.
   2. Fix it with the smallest change that is safe.
   3. Run the project once more and confirm the app is working.
   
   Do not touch files unrelated to the problem unless the fix requires it.

4. Phase 4 — Add one featureCopy

   Next feature to add: [feature]
   
   - Before you write code, tell me which files you will modify and why.
   - Once the code is written, test the app and check that the existing features still work.
   - Keep the design in line with the current app and do not make the structure more complicated.

5. Phase 5 — Product-design reviewCopy

   Look at this app the way a product designer would and tell me:
   - what feels confusing
   - what should be taken out
   - what should be made clearer
   - what a person using it for the first time would miss
   - what would make the app more useful
   
   Then improve the interface without making it complicated. It must stay clean, simple and easy to understand.

6. Phase 6 — Launch checklistCopy

   Before launch, draw up a checklist for this app. Look for:
   - broken pages
   - errors in the console
   - layout problems on mobile
   - copy that is confusing
   - missing empty states and loading states
   - basic accessibility issues
   - anything else that makes the app feel unfinished
   
   Sort the issues from most to least important, then fix the top ones first.