spec-driven-development

To turn a vague idea into an executable specification, you provide your project concept and desired outcomes to a structured workflow that enforces frontmatter and generates implementation-ready PRDs, architecture notes, and testing plans.

Creating a PRD for AI coding agents generates three core artifacts: spec.md for the detailed specification, prompt_plan.md for guiding AI prompts, and todo.md for a structured task checklist.

Yes, this specification workflow applies to both greenfield and existing projects, enforcing structured frontmatter and phased workflows to ensure clear requirements and measurable success criteria across any codebase.

Writing implementation-ready specifications involves a three-phase workflow: Idea Honing to refine concepts, Planning to establish architecture notes and success criteria, and Execution to generate detailed plans and task checklists for AI agents.

No specific dependencies are required to generate structured prompt plans for AI development. The workflow operates independently to transform project ideas into clear instructions and testing plans.

Using a phased workflow for PRD creation ensures vague ideas are systematically refined into measurable success criteria and structured frontmatter, preventing misalignment between project requirements and AI agent execution outputs.