Complementary Perspective Panel

Own a single-response reasoning structure built from three complementary method lenses. It is not a claim that real experts were consulted and is not an execution topology.

Route

Use when a systemic problem genuinely spans distinct disciplines, stakeholder systems, or failure modes and the user needs synthesis rather than one expert answer. Select three lenses that change what is noticed, not three titles that repeat the same viewpoint.

Skip simple domain questions, ordinary brainstorming, requests for an actual named expert’s view, or a single pro/con dispute that $bidirectional-steelman can own. Route screenplay diagnosis to $script-doctor-roundtable. Use $agent-harness-review-loop only when real multi-agent execution is separately authorized.

Procedure

1. Restate the problem, known facts, goal, constraints, and unresolved decision.
2. Choose three complementary lenses and state why each is necessary.
3. For each lens, provide: its reframing, recommended route, risk the other lenses may miss, and evidence that would change its view.
4. Cross-question the lenses to find shared facts, genuine branch points, and the different assumptions behind disagreement.
5. Synthesize a recommendation, applicability conditions, largest risk, exit condition, and first action.

Handoffs and Boundaries

• Never impersonate or fabricate the opinion of a real person.
• Do not spawn agents merely because three perspectives are requested; multiple agents and majority voting are not evidence.
• Use $fact-claim-audit for shared external facts.
• Add $bidirectional-steelman only when the synthesized plan still contains one material conclusion-flipping choice.

Completion Gate

The three lenses are materially different, each contributes a unique risk or constraint, disagreements trace to explicit assumptions, and the final synthesis includes conditions, risk, exit rule, and first step.

Source Status

Distilled from the user-supplied “专家会诊” method card. The card proposes a multi-perspective prompt; it does not document real expert involvement or outcome evidence.