tlc-spec-driven

To turn natural-language requirements into traceable specifications, the Skill converts your intent into structured specs with unique IDs (e.g., AUTH-01) that flow through design, task breakdown, and verification. It skips ceremony for small changes while maintaining full traceability for complex features.

The best way to break down features into verifiable tasks is using an adaptive pipeline that generates tasks with explicit 'Done when' criteria. This ensures each task is verifiable and executed with atomic git commits, scaling the process based on feature complexity.

Yes, this spec-driven workflow is stack-agnostic and works across any coding agent and technology stack. It applies to new project initialization, brownfield codebase mapping, and feature implementation without requiring specific frameworks or platforms.

Yes, you can pause and resume feature implementation across sessions using the built-in session continuity support. It provides handoff capabilities and maintains persistent state, including decisions, blockers, and deferred ideas, ensuring smooth context restoration.

To map an existing codebase before planning, the Skill performs brownfield analysis to generate mapping documentation. This allows you to understand the current architecture and seamlessly integrate new feature specifications into the existing project structure.

You should skip the design phase during feature task breakdown for small or straightforward changes that affect three files or fewer. The adaptive planning pipeline automatically skips ceremony, moving directly from specification to execution for quick fixes.