Overview

The skill asks Claude to model a decision before recommending an answer. It separates options from conditions, identifies missing evidence, and makes the stopping rule explicit.

Workflow

  1. State the decision and the outcome being optimized.
  2. List viable options without forcing a preference too early.
  3. Convert assumptions into branch conditions.
  4. Identify the evidence required to enter each branch.
  5. Set the stopping rule: when to act and what would reopen the decision.
Do we have enough evidence?
YesChoose the smallest safe action
NoRun one evidence-gathering test

Good use cases

  • Choosing which product direction to test.
  • Deciding whether to automate or keep a manual process.
  • Evaluating a tool when cost, risk, and learning time conflict.
  • Defining escalation rules for a recurring operational decision.

Limits

A decision tree makes logic inspectable; it does not make weak evidence stronger. For irreversible or high-stakes decisions, domain expertise still matters.