What problem does it solve?
It solves the problem of turning vague AI coding sessions into a disciplined, measurable engineering process with explicit artifacts, validation, and traceable decision-making.
Core Features & Use Cases
- Stage-Orchestrated Engineering Workflow: Coordinates the full six-stage (Discover → Design → Demonstrate → Develop → Document → Debrief) or fast three-stage (Plan → Build → Verify) progression.
- Artifact and State Management: Tracks active stage, writes stage outputs to the correct
.dew/ paths, and advances/records progress in .dew/state.md.
- Context Loading and Handoff to Stage Skills: Loads prerequisite artifacts per stage and invokes the appropriate stage skill so progress is preserved within a session.
- Git/Worktree Boundary Discipline: Optionally commits stage boundaries based on
.gitignore rules and enforces a worktree policy to keep main clean.
Quick Start
Start a new dew project by running /dew new and answering the prompts for project name and whether you want the full or fast workflow.