project-init

To scaffold an AI-ready project, this skill generates CLAUDE.md, a Diátaxis documentation tree, and datarim workflow state directories. It conditionally fills build placeholders using a separate tech-stack selection document to ensure consistent AI-assisted execution.

Idempotent project scaffolding initializes a project structure without overwriting existing files. It only creates missing directories and reports skipped files, allowing safe incremental re-scaffolding when project intent signals change during AI-assisted workflows.

Setting up a Diátaxis documentation taxonomy is handled by generating a standardized documentation tree during project initialization. This creates a structured layout for tutorials, how-to guides, references, and explanations tailored for AI execution.

Yes, you can re-scaffold an existing project without losing files because initialization operations are idempotent. The process detects project intent signals via triggers, only creates missing directories or files, and explicitly reports what was skipped.

Yes, project initialization generates a stack-appropriate .gitignore. It reads your tech stack configuration from a separate decision document and fills the CLAUDE.md build and config placeholders accordingly to match your stack.

Workflow state directories are needed to maintain consistent execution context across AI-assisted development sessions. They store datarim workflow states, preventing repeatability issues and missing state tracking when starting new projects.