design-solution

Design Solution Skill

[!IMPORTANT] Turn an approved PRD or implementation goal into SRS/FRS technical requirements (How), architecture, contracts, and verification decisions.

Optional args: slug=, ticket=<id/url>, mode=interactive|autonomous|channel, channel=, auto_continue=true|false, profile=business|hybrid|technical.

Instructions

When the user asks to perform this workflow, execute the following steps:

Design Solution Workflow (SRS/FRS / How)

Goal: Produce a build-ready technical design with explicit boundaries, contracts, risks, and tests.

Steps

1. Load inputs:
   • Load baseline SRS/FRS section, common-software-requirements, common-decision-discipline (option cards, evidence ledger), PRD or ticket, implementation plan, matched framework skills, architecture docs, trace source BRD-OBJ-* -> REQ-* -> AC-*, and product-management/product-requirements/references/requirements-modeling.md from the owning package when modeling guidance applies; report excluded or unavailable resources without guessing.
2. Define architecture:
   • Name bounded contexts, module/data owners, migration needs; select technical contracts/views without inventing product scope.
   • Consume or create HLD trace: requirements, audience, scope, shaping constraints, lifecycle status, ownership, failure domains, decisions.
   • Define dependency direction, component RACI, and sync, async, or hybrid communication.
   • Define early mock/schema contracts for parallel start; create LLD contracts only for components/flows needing them. Select container, sequence, or other views by question through common-architecture-diagramming or requirements-modeling.md (evidence cites SRS/system-design doc, metric from stated NFR thresholds). Views clarify interaction/lifecycle/order; diagrams not mandatory when prose or tables are precise enough; never fabricate metrics.
3. Define contracts:
   • Functional flows (FRS): user/system steps, inputs/outputs, validations, and error states.
   • For complex flows, use one actor, one goal, one session; split normal course from alternatives and exceptions.
   • Requirement cards: statement, priority, status, source, behavior, NFRs, measurement, and verification lane.
   • API inputs/outputs and interface contracts (OpenAPI/Protobuf).
   • Events/jobs and async guarantees (at-least-once, idempotent).
   • Storage shape, ownership, retention, and migration rules.
   • Security, permission, and privacy checks.
   • Carry REQ-* -> HLD-* -> CMP-* -> LLD-* -> VER-* IDs into requirement cards and interface contracts. LLD is the low-level-design lane: specify ownership, consistency, ordering/idempotency, adverse timeline, recovery, and verification for each chosen component; do not force every component into a diagram.
   • NFR thresholds for performance, reliability, and scalability.
4. Plan verification:
   • Unit, integration, E2E, visual, mobile, security, and migration checks.
   • Failure mode analysis for dependencies, fallbacks, retries, and rollback/degradation.
   • Save technical requirements to docs/srs/srs-[slug].md when file writes are allowed.
   • Record evidence in docs/srs/srs-walkthrough-[slug].md.
5. Record ADR:
   • Write one concise ADR when architecture or public contract changes.
   • Continue when patterns are inferable; return BLOCKED for cross-team contracts, migrations, permissions, or NFR uncertainty.
   • Route next step to implementation-readiness or dev-fix.