Enforce design validation before implementation

Design Orchestration is a meta-skill that ensures ideas become designs, designs are reviewed, and only validated designs reach implementation. It does not generate designs. It is a routing and enforcement skill, not a creative one, and it controls the flow between other skills by deciding which skill must run next, whether escalation is required and whether execution is permitted. It coordinates brainstorming for design generation, multi-agent-brainstorming for design validation, and downstream implementation or planning skills.

The routing logic runs in five steps:

• Step 1, brainstorming (mandatory): if no validated design exists, invoke brainstorming and require an Understanding Lock, an Initial Design and a started Decision Log. You may not proceed without these artifacts.
• Step 2, risk assessment: classify the design as low, moderate or high risk, using factors such as user impact, irreversibility, operational cost, complexity, uncertainty and novelty.
• Step 3, conditional escalation: low risk proceeds to implementation planning, moderate risk gets a recommendation for multi-agent-brainstorming, and high risk requires it. Skipping a required escalation is prohibited.
• Step 4, multi-agent review, if invoked: it requires the completed Understanding Lock, the current design and the Decision Log. It allows no new ideation, no scope expansion and no reopening of the problem definition, so only critique, revision and decision resolution take place.
• Step 5, execution readiness check: before implementation, confirm the design is approved by a single agent or by multiple agents, the Decision Log is complete, major assumptions are documented and known risks are acknowledged. If any condition fails, block execution and return to the appropriate skill.

Enforcement rules: do not allow implementation without a validated design, do not allow skipping required review, do not allow silent escalation or de-escalation, and do not merge the design and implementation phases.

The skill exits only when the next step is explicitly identified and all required prior steps are complete. The possible exits are to proceed to implementation planning, run multi-agent-brainstorming, or return to brainstorming for clarification. If a reviewed design reports a final disposition of APPROVED, REVISE or REJECT, the workflow must be routed accordingly with the chosen next step stated explicitly.

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. Do not use it to create the design itself, because it has no creative role. Use it only when the task clearly matches this scope. Its output is not a substitute for environment-specific validation, testing or expert review, and it stops to ask for clarification if inputs, permissions, safety boundaries or success criteria are missing.

The input is a proposed feature, system or change, plus the artifacts produced by the design skills: the Understanding Lock, the design and the Decision Log. The output is an explicit routing decision naming the next step, or a block on execution that sends the work back to the right skill. An example request is to review an interface with the skill, identify the highest-impact design problems and propose an implementation-ready improvement.

Teams and agent operators who want a gate between ideation and code, so risky designs get reviewed before anything is built. The stated philosophy is to slow down the right decisions, speed up the right execution and prevent costly mistakes: good systems fail early and bad systems fail in production.