Put one situation in front of three professionals chosen to disagree — the lawyer, the accountant, and the therapist read the same divorce — each ruling independently, with the conflicts surfaced in a disagreement table instead of blended into mush. Use when asked for different professional perspectives, what would a lawyer versus an accountant say, I keep getting one-sided advice, or pressure-test this from multiple angles. Produces three independent readings, the disagreement table with why the frames diverge, and the decision map of which expert to weight for which sub-decision. The anti-consensus skill: it never averages.
▸The situation — in full, messy is fine; each expert will extract what their frame considers relevant, and the differences in extraction are part of the output
▸The decision in play — what actually gets decided after this, so the readings aim at something
▸The panel, if the requester wants to choose — otherwise the skill selects for disagreement and says why; a panel that will agree is a wasted panel
▸The advice already received — whose frame has already been heard, so the panel can be weighted toward the unheard frames
▸The constraints — anything already fixed (deadlines, budgets, relationships that must survive), which the experts must respect rather than advise away
✅ The bar it holds itself to
Every skill in this library self-verifies — these are this skill's own quality checks, straight from its definition.
✓The panel selection predicts the disagreements before the readings exist
✓Each reading stands alone, in its own register, with its bias intact and labelled
✓The disagreement table explains *why* frames diverge, not just that they do
✓Each conflict row says whether information or values would settle it
✓The convergence list exists and is genuinely actionable
✓The decision map assigns experts to sub-decisions instead of crowning one
⚠️ What it refuses to do
**The blend.** Any sentence beginning "on balance, the experts suggest" has destroyed the product; the conflicts are the deliverable.
**The strawman seat.** Three experts where one exists to lose is a rigged panel; every seat must be capable of being right.
**De-biasing the professionals.** The accountant who also weighs the emotional side has stopped being the accountant; the panel already has a therapist.
**Settling values rows with fake facts.** "It depends what you value more" is an honest table entry, and pretending data settles it is the consensus disease returning.
**Panel sprawl.** Four-plus experts turns readings into a survey; three frames in real tension is the format.
Install
npx pm-claude-skills add --agent claude # or codex · cursor · gemini · hermes
# or one-line MCP (every skill, any client):
claude mcp add pm-skills -- npx -y pm-claude-skills-mcp