universal-app-autobuilder

Prompt file imported from ahmetcemkaraca/TourTipApp (.github/prompts/universal-app-autobuilder.prompt.md). Fill in {{Write_your_app_idea_or_needs_in_one_paragraph}} before use. Copyright stays with the author.

Input

• {{Write_your_app_idea_or_needs_in_one_paragraph}}

Context

• Guide: ../instructions/universal-agent.instructions.md
• Roles: ../instructions/architect.instructions.md, ../instructions/developer.instructions.md, ../instructions/qa.instructions.md, ../instructions/security.instructions.md, ../instructions/devops.instructions.md, ../instructions/ux.instructions.md
• Project memory: requirements.md, design.md, tasks.md at the workspace root (create if missing)

Goals

1. Analysis: scope, risks, open questions (short)
2. Trio: requirements.md (EARS + AC + NFR), design.md (architecture + diagrams), tasks.md (ordered tasks)
3. Implementation: first vertical slice (UI+API+DB or platform equivalent) + tests
4. Run: local run/test commands and expected outputs
5. Ship: minimal deployment (web: Vercel/Netlify or Docker; mobile: Expo dev client/TestFlight; desktop: packaging notes)
6. Quality gates: lint+tests green; fast fixes on failures
7. ToDo: short list of next steps

Rules

• Ask questions only if truly blocking; otherwise make reasonable, reversible assumptions.
• Proceed in order: Plan → Validate → Code → Test → Ship.
• No secrets in code; document with .env.example and use env vars.
• Keep chat replies short; provide complete, runnable file changes.

Coaching and Alternatives

• If you detect wrong/high-risk assumptions, mark them as “Incorrect assumption”, briefly explain why, and offer 2–3 safe/applicable alternatives.
• For each alternative: summarize expected benefit, risk, cost/time, and complexity in 1–2 sentences; choose a “recommended path”.
• Explain technical terms briefly in plain language (parenthetical notes); include sample commands or tiny code snippets when helpful.
• For dangerous/inappropriate requests, propose safer equivalents and why; suggest staged rollout (feature flag) when relevant.

Output Format

• Sections: Analysis → Document updates → Code changes → Commands → Test/Smoke results → Deploy steps → Next TODO
• List changed files; put commands on separate lines.

Edge Cases and Special Scenarios

• If the repo is empty: scaffold the chosen stack (Next.js/Electron/Expo/Flutter, etc.) and add one vertical slice.
• Mobile/desktop: only document what’s necessary for signing/deployment; do not request secret keys.
• If external services are missing: mock behind an interface and include explicit setup instructions.

Start

• Summarize the user input; produce/update the trio; implement the first vertical slice; write commands and deploy steps; finish with a TODO list.