df:new-project

To start a new project, this workflow automates initialization by guiding context-gathering questions and generating planning artifacts like ROADMAP.md and REQUIREMENTS.md under a .planning directory. It establishes a repeatable startup process with gated validation steps.

Project initialization with deep context engineering is a process that uses guided questioning and optional research to gather project scope and constraints, then generates structured planning materials to maintain project memory and ensure a repeatable startup workflow.

Yes, you can use this project initialization workflow without external dependencies. It operates independently with frontmatter-driven configuration to generate planning artifacts and manage end-to-end workflow gates, requiring no prerequisite components.

The best way to maintain project state and memory during setup is by automatically generating a STATE.md file within a dedicated .planning directory, which preserves context and tracks workflow progress alongside other planning artifacts.

The workflow handles validation and approvals by preserving end-to-end workflow gates throughout the startup process, ensuring that all planning artifacts and project contexts are validated and approved before proceeding to subsequent steps.

You need a dedicated .planning directory for project artifacts when initializing a new project that requires structured planning materials like PROJECT.md, ROADMAP.md, and STATE.md to establish clear goals, scope, and a repeatable startup workflow.