What problem does it solve?
Prevents AI coding agents from rushing into implementation by forcing a disciplined, spec-before-code feature lifecycle that reduces rework, scope creep, and undocumented behavior.
Core Features & Use Cases
- Feature Definition with gated checkpoint: Converts a feature request into a bounded scope and initializes a plan, then waits for user acknowledgment before progressing.
- Behavioral Spec Gate and architecture review: Ensures a reviewed, complete spec exists and that internal architecture decisions are explicitly approved before any code changes.
- BDD-driven development with quality baselines: Runs lint/type/audit checks on existing files, validates external API assumptions, writes failing BDD tests from the spec, implements to satisfy those tests, and verifies 100% coverage with an explicit plan closure.
- Prerequisite workflow spawning: Routes into
refactor-workflow when structural prerequisites are required, and resumes the feature workflow after the prerequisite merges.
Quick Start
Use the feature-workflow skill when the user asks for a new feature or enhancement so the agent first defines scope, confirms a reviewed spec, then designs, tests, implements, and verifies coverage before closing the plan.