What problem does it solve? Designing systems without a disciplined process leads to decorative boundaries, leaked contracts, unplanned failure modes, and specs that fall apart when executed by someone with less context. This Skill provides a structured design methodology so architecture decisions are explicit, evaluated, and survivable. ## Core Features & Use Cases - Design Loop: Model the forces first, define boundaries as contracts with violation signals and owners, and design every flow for mid-step process death with recovery paths. - Decision Discipline: Classify decisions as reversible versus expensive to reverse, generate at least two approaches, run pre-mortems, and record significant choices as Architecture Decision Records using the included ADR reference. - Decomposition & Specs: Split work into independently verifiable parts ordered by risk, and write specs (role, what, how) executable by a colder, cheaper context. - Use Case: When planning a multi-component feature, use this Skill to enumerate consumers of shared surfaces, compare two architecture options, document the tradeoff as an ADR, and hand off specs that another agent can execute without your context. ## Quick Start Ask the agent to design the architecture for a new multi-component feature, defining boundaries, contracts, failure recovery, and an ADR for the key tradeoff.
