App Starter Prompts

A collection of structured AI prompts for kicking off new software projects. Use these as starting points when working with an AI coding assistant to produce a complete, implementation-ready spec.

How to use

1. Pick the prompt that matches your situation (see table below)
2. Copy the entire file contents
3. Paste it into your AI assistant as your first message
4. Answer the staged questions — the AI guides you through them one stage at a time
5. At the end, the AI generates the output documents
6. Save those files into your project's /docs folder and start building

Prompts

File	Purpose
docs/00-kickoff-web-app.md	Web / frontend project kickoff — produces MVP_DOCUMENT.md, DECISIONS.md, OPEN_QUESTIONS.md
docs/00-kickoff-api.md	API / backend project kickoff — produces API_DOCUMENT.md, DECISIONS.md, OPEN_QUESTIONS.md
docs/01-feature-spec.md	Plan a new feature in an existing project — produces FEATURE_SPEC.md, IMPLEMENTATION_PLAN.md
docs/02-claude-md.md	Generate a CLAUDE.md quick-reference for an existing codebase
docs/03-roadmap.md	Plan phased rollout and track build progress — produces ROADMAP.md, BUILD-STATUS.md
docs/04-design.md	Generate a reusable design system prompt for AI design tools — produces DESIGN_PROMPT.md

Examples

Finished sample outputs live in examples/:

• examples/00-kickoff-web-app/ — sample MVP_DOCUMENT.md, DECISIONS.md, OPEN_QUESTIONS.md
• examples/00-kickoff-api/ — sample API_DOCUMENT.md, DECISIONS.md, OPEN_QUESTIONS.md
• examples/01-feature-spec/ — sample FEATURE_SPEC.md, IMPLEMENTATION_PLAN.md
• examples/02-claude-md/ — sample CLAUDE.md
• examples/03-roadmap/ — sample ROADMAP.md, BUILD-STATUS.md
• examples/04-design/ — sample DESIGN_PROMPT.md

Use them to calibrate prompt changes and check whether generated outputs are concrete enough to implement.

Maintenance

• docs/ is the source of truth for prompt behavior and required output sections.
• examples/ are canonical reference outputs and should stay structurally aligned with the current prompts.
• If a prompt's output contract changes, update the corresponding files in examples/ in the same PR.
• Run ./scripts/validate.sh after changing docs/ or examples/.

Output documents

Kickoff prompts (00-kickoff-*.md) produce:

• MVP_DOCUMENT.md / API_DOCUMENT.md — complete implementation spec (pages, schema, endpoints, security, project structure, env vars, build order)
• DECISIONS.md — architectural decision records (ADRs) with alternatives and rationale
• OPEN_QUESTIONS.md — unresolved questions, edge cases, and post-v1 ideas

Feature spec (01-feature-spec.md) produces:

• FEATURE_SPEC.md — complete feature spec (flows, UI, API/data changes, permissions, edge cases, observability, rollout)
• IMPLEMENTATION_PLAN.md — task breakdown with acceptance criteria, dependencies, and risk table

CLAUDE.md generator (02-claude-md.md) produces:

• CLAUDE.md — AI quick-reference: stack, commands, structure, conventions, env vars, do-nots

Roadmap & build status (03-roadmap.md) produces:

• ROADMAP.md — phased plan with decision log, non-goals, success criteria, and exit gates per phase
• BUILD-STATUS.md — live workstream tracker pre-populated from the spec, with source-of-truth pointers and task lists

Design system prompt (04-design.md) produces:

• DESIGN_PROMPT.md — master design prompt plus page-specific and component variant prompts for AI design tools (v0.dev, Bolt, Lovable, Stitch, Galileo)