prompt-export_skill

Overview

This skill converts completed design artifacts—problem framing, PRD, and UX specs—into a sequenced prompts.md file that guides Claude Code through incremental development. It bridges design thinking to executable, checkpointed prompts so coding agents stay focused and produce verifiable increments. Use it after upstream design work is done to produce a clear developer-facing playbook.

How this skill works

Provide your design artifacts (markdown, docs, images, or bullet notes). The skill extracts project overview, core features, user flows, component descriptions, interactions, constraints, tech preferences, and success criteria. It then fills gaps with a few targeted questions and decomposes the project into a logical sequence of self-contained prompts, each with a goal and checkpoint, and outputs a prompts.md file for Claude Code.

When to use it

• You have completed problem framing, PRD, and UX specs and need actionable developer prompts.
• You want to convert product and UX artifacts into step-by-step tasks for a coding agent.
• You need to enforce checkpoints so an agent doesn’t skip assumptions or lose focus.
• You’re preparing handoff material for iterative, agent-driven implementation.
• You want prompts that reference specific files and include acceptance criteria.

Best practices

• Supply complete artifacts up front; include flows, screens, and any tech preferences.
• Answer minimal clarification questions to resolve critical gaps (tech stack, auth, persistence).
• Keep prompts focused and single-purpose with clear acceptance criteria and file references.
• Choose granularity based on complexity: simple (3–5) to complex (10+) prompts.
• Work sequentially through prompts and verify checkpoints before advancing.

Example use cases

• Turn a PRD and wireframes into a 7-step prompts.md for a React MVP with Tailwind.
• Convert mobile-first UX specs into a sequenced plan that builds screens, state, and auth.
• Generate integration and polish prompts for an app where core components are already scaffolded.
• Create a phased prompt sequence for a complex product with dependencies and rollback hints.