Enforce design validation before implementation

Design Orchestration is a meta-skill that doesn't generate designs itself - it routes and enforces the flow between other design skills, so that ideas become designs, designs get reviewed, and only validated designs reach implementation. It coordinates brainstorming (design generation) and multi-agent-brainstorming (design validation), deciding which skill runs next, whether escalation is required, and whether execution is permitted. When no validated design exists, it mandates running brainstorming first and requires an Understanding Lock, an Initial Design, and a started Decision Log before proceeding. It then classifies the resulting design as low, moderate, or high risk based on user impact, irreversibility, operational cost, complexity, uncertainty, and novelty: low risk proceeds straight to implementation planning, moderate risk recommends multi-agent-brainstorming, and high risk requires it - skipping a required escalation is prohibited. If multi-agent review runs, it must only critique, revise, and resolve decisions - new ideation, scope expansion, or reopening the problem definition are not allowed. Before implementation is permitted, it confirms the design is approved, the Decision Log is complete, major assumptions are documented, and known risks are acknowledged; failing any of these blocks execution and routes back to the appropriate skill.

Invoke it when a user proposes a new feature, system, or change, when a design decision carries meaningful risk, or when correctness matters more than speed. It is not itself a creative skill and won't produce a design - use brainstorming or multi-agent-brainstorming directly for that; this skill only governs when those skills run and whether their output is sufficient to move forward.

Input: a proposed feature, system, or change, plus the current state of any Understanding Lock, Design, and Decision Log artifacts. Output: an explicit routing decision - proceed to implementation planning, run multi-agent-brainstorming, or return to brainstorming for clarification - or, once a reviewed design reports APPROVED, REVISE, or REJECT, the corresponding next step stated explicitly. It never merges the design and implementation phases and never allows silent escalation or de-escalation.

• brainstorming skill for design generation
• multi-agent-brainstorming skill for design validation and review
• Downstream implementation or planning skills, gated on execution-readiness checks

Its stated design philosophy is to slow down the decisions that matter and speed up execution once those decisions are made, on the premise that good systems fail early in review rather than in production.

Teams or agent workflows that want a hard gate preventing implementation from starting before a design is generated, risk-assessed, and, where risk warrants it, independently reviewed.