# Product Decision Alignment > Diagnose and repair a stuck, repeatedly reopened, or inconsistently executed cross-functional product decision. Use when product, design, engineering, operations, or leadership disagree about the outcome, evidence, authority, decision method, commitment, or follow-through. Not for interpersonal therapy, performance evaluation, or choosing the product strategy itself. - Skill: `majesticlabs-dev/product-decision-alignment` (Agent Skill) - Install (CLI): `npx skillmds@latest add majesticlabs-dev/product-decision-alignment` - Raw SKILL.md: https://api.skillmd.com/api/skills/majesticlabs-dev/product-decision-alignment/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: majesticlabs-dev (https://skillmd.com/u/majesticlabs-dev) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/majesticlabs-dev/product-decision-alignment --- # Product Decision Alignment Make the decision structure explicit. Alignment means people understand the decision, authority, rationale, obligations, and dissent. It does not require enthusiasm or unanimous agreement. ## Boundary Use this skill for one consequential product decision or a tightly related decision set. Do not infer motives, trust, competence, or bad faith from disagreement. Do not infer agreement from silence, meeting attendance, or lack of public objection. A decision can be clear while commitment, feasibility, or legitimacy remains unresolved. This skill diagnoses the decision system and recommends a repair. It does not mediate personal conflict, make an executive decision without authority, or manufacture consensus. ## Required Inputs Establish: - exact decision, current status, deadline, and cost of delay - intended product or business outcome - affected groups and implementation owners - evidence, assumptions, constraints, and disputed facts - decision owner, delegated authority, veto or escalation rights - decision method and consultation expectations - prior decisions, reversals, and current commitments Use `Supported`, `Contradicted`, or `Unknown` for each diagnostic conclusion. ## Workflow ### 1. Write the decision sentence State what must be decided, by whom, by when, for which scope, and what remains outside the decision. Separate the decision from discussion topics and implementation details that can be delegated. ### 2. Map participation and authority Identify: - decision owner - people providing evidence or specialist judgment - people responsible for implementation - people materially affected - valid approval, veto, or escalation authority - people who need notification rather than decision power Flag overlapping or missing authority. Do not solve authority ambiguity by adding everyone as an approver. ### 3. Assess six alignment conditions Review each independently: 1. **Outcome:** Is the intended result and success evidence shared? 2. **Evidence:** Are facts, assumptions, uncertainties, and disputed claims visible? 3. **Interests:** Are incentives, constraints, risks, and accepted tradeoffs explicit? 4. **Authority:** Is it clear who decides and where that authority comes from? 5. **Method:** Is the decision process, consultation window, and deadline understood? 6. **Commitment:** Is the decision recorded with owners, obligations, dissent, and revisit conditions? Several conditions may fail at once. Do not call the first failure the universal root cause. ### 4. Distinguish disagreement types Classify the material disagreement as one or more of: - different desired outcomes - different interpretation of evidence - different risk tolerance or incentive - disputed authority - unclear decision method - unresolved feasibility or capacity - unrecorded decision - non-commitment after a clear decision - new evidence that legitimately reopens the choice State the evidence for the classification and competing explanations. ### 5. Choose the smallest repair Match the intervention to the failure: - rewrite the outcome and success measure - resolve a factual dispute with a named evidence task - surface a tradeoff or incentive conflict for owner judgment - clarify or escalate decision authority - time-box consultation and define the decision method - record the decision, owner, rationale, obligations, and dissent - establish implementation checkpoints - define evidence that permits reconsideration When decision authority is absent or disputed, report it as unresolved and use the valid escalation path. If work must coordinate temporarily while authority is resolved, a visible convention or precedent may support a reversible coordination proposal, but it does not constitute an approved decision. Label its owner, duration, limits, and rollback condition. Before recommending it, check legality, safety, affected-party impact, reversibility, and whether a better-authorized process is available. Never create convergence through deception, manufactured urgency, or hidden irreversible action. ### 6. Close the loop Specify who communicates the decision, who acknowledges implementation obligations, where the record lives, when progress is checked, and what exact evidence can trigger revisit or escalation. ## Output 1. **Decision sentence and status** 2. **Participant and authority map** 3. **Alignment assessment** with Supported, Contradicted, or Unknown evidence 4. **Disagreement diagnosis** and competing explanations 5. **Immediate repair and owner** 6. **Decision record template** 7. **Commitment, follow-through, and revisit rules** Default to one compact condition-and-authority matrix, at most three material causes, and the smallest matching repairs. Expand only when the decision has several genuinely independent conflicts. ## Quality Gate - The analysis concerns a specific decision, not vague organizational health. - Facts, assumptions, interests, and authority are not conflated. - Silence is not treated as agreement. - Multiple causes can remain active. - The repair is smaller than the dysfunction it addresses. - Dissent remains visible after the decision. - Revisit rules require changed evidence or conditions, not political persistence.