# Decision Interview Builder > Conduct a structured multi-turn interview through one real past decision the user made — walking situation, decision, risk, and change — and produce a sanitized, shareable judgment artifact that documents the reasoning without leaking confidential details. Use when the user says "interview me about a decision", "build my judgment artifact", "extract my reasoning", "decision interview", "show my thinking", or wants to generate career evidence from a real work decision. - Skill: `m2ai-portfolio/decision-interview-builder` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add m2ai-portfolio/decision-interview-builder` - Raw SKILL.md: https://api.skillmd.com/api/skills/m2ai-portfolio/decision-interview-builder/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: m2ai-portfolio (https://skillmd.com/u/m2ai-portfolio) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/m2ai-portfolio/decision-interview-builder --- # Decision Interview Builder Turns one real past decision into a portable, sanitized judgment artifact. The output travels with the person — surviving job changes, badge swipes, and resume rewrites — because it proves *how* they thought, not just *what* they shipped. ## Trigger Use when the user says "decision interview", "interview me about a decision", "build my judgment artifact", "extract my reasoning", "show my thinking", "prove I understood that call", or wants documented evidence of their reasoning for a career file, promotion packet, or interview prep. ## Phase 1: Decision Selection Ask the user to name one real decision they have made. Accept any level: - An IC choosing between two technical approaches - A team lead deciding whether to ship or delay - A manager making a hiring call under uncertainty - An executive choosing between two market bets If the user gives a vague situation rather than a decision (e.g., "we had a big project"), redirect: "Tell me the specific choice point — what were the two or more options on the table?" Do not proceed until there is a clear decision with at least two alternatives. ## Phase 2: Four-Question Interview Walk the user through the four questions in sequence. One question at a time — do not batch them. After each answer, probe once for specificity if the answer is vague (see below). ### Q1: Situation "Describe the situation you were in when this decision needed to be made. What was at stake? Who was affected?" Probe if vague: "What would have broken or gone wrong if no decision had been made at all?" ### Q2: Decision "What did you actually decide? Walk me through the moment of choosing — what were the real options, and what tipped you toward this one?" Probe if vague: "What was the option you *didn't* choose, and what would have happened if you had?" ### Q3: Risk "What was the biggest risk you accepted by making this call? What did you know you might lose or break?" Probe if vague: "If this decision had backfired completely, what would that have looked like three months later?" ### Q4: Change "What changed as a result of this decision — in the outcome, in how you would approach the same situation again, or in what you now know you were wrong about?" Probe if vague: "What is the one thing you would tell yourself before this decision that you only know now?" ## Phase 3: Sanitization Pass Before drafting the artifact, ask: "Before I write this up, are there any details in your answers — company names, product names, client names, colleague names, specific numbers — that you'd like me to replace or generalize?" Replace any flagged details with generic labels: [company], [client], [Product X], [Q3], [team lead], etc. Keep the reasoning structure intact. ## Phase 4: Artifact Output Produce a clean, portable judgment artifact in the following format: ``` # Judgment Artifact — [Decision Title, e.g., "Build vs. Buy: Auth System"] ## Situation [2–4 sentences: the context, what was at stake, who was affected] ## Decision [2–3 sentences: what was chosen, the real alternatives, what tipped the call] ## Risk Accepted [2–3 sentences: the known downside, what could have broken] ## What Changed [2–3 sentences: the actual outcome or ongoing result, what the decision-maker now carries forward] ## Judgment Signal [1 sentence: the non-obvious insight this decision demonstrates about how the person thinks — written in third person, suitable for a promotion packet or interview answer] ``` The **Judgment Signal** line is the most important. It is not a summary — it is the distillation of the decision-making quality embedded in the story. Examples: - "Shows ability to ship under resource constraint without compounding technical debt downstream." - "Demonstrates willingness to reverse a sunk-cost position when new information changed the risk calculus." - "Identifies when stakeholder alignment is the bottleneck, not the technical solution." ## Phase 5: Storage Prompt After delivering the artifact, ask: "Where would you like to store this? Options: paste into your own document, save to a local file, or add to a decision log you're building." Do not hardcode any path. Offer to write to `./decision-artifacts/[slug].md` if the user is in a project directory; otherwise default to displaying for copy-paste. ## Verification - [ ] The decision has at least two named alternatives (not just "we did X") - [ ] The Risk section names a real downside that was accepted, not just "it could have failed" - [ ] The What Changed section includes something the decision-maker *learned* or would do differently - [ ] The Judgment Signal is in third person and is specific enough to stand alone in a promotion packet - [ ] No confidential detail survived to the artifact without the user's explicit approval ## Source Derived from Nate's Newsletter 2026-05-31, "Executive Briefing: Your career evidence is thinner than you think + 3 prompts that rebuild it." Nate's "builder that interviews you through one real decision" is the core concept; this skill operationalizes it as a structured four-question interviewer with a sanitization pass and a portable artifact output.