What problem does it solve? When multiple developers or agents build different parts of a system independently, they make incompatible choices about boundaries, state mutation, data ownership, and dependencies. This Skill produces an architecture spine — a lean contract of invariants (AD-n decision blocks with Binds/Prevents/Rule) that fixes only the non-obvious, load-bearing calls so everything built from it stays consistent. ## Core Features & Use Cases - Architecture Spine Authoring: Distills decisions into a terse ARCHITECTURE-SPINE.md with stable AD IDs, consistency conventions, a pinned stack table, mermaid structural diagrams, and an explicit Deferred section. - Coaching or Fast Path: Elicits decisions through open-ended coaching by default, or drafts the full spine with [ASSUMPTION] tags for fast review. - Brownfield & Inherited Spines: Ratifies conventions from an existing codebase, and treats a parent spine's ADs as binding read-only constraints when working at epic altitude. - Reviewer Gate: Runs a deterministic lint_spine.py pass plus parallel reviewer subagents against a good-spine checklist before handoff; also supports a standalone Validate intent with an HTML report. - Use Case: Given a spec package for a new platform, run the skill to produce an architecture spine fixing the design paradigm, dependency rules, and shared-data ownership, then hand it to epic and story breakdown workflows. ## Quick Start Ask the AI to create the technical architecture for your project from your spec or existing codebase, choosing the coaching path to work through the key decisions together.
