Critical Thinking Workflow
7-skill end-to-end process: reframe → surface assumptions → clean reasoning → downstream effects → calibrate confidence → make tradeoffs explicit.
Workflow Overview
[1] Question Upgrader
|
v
[2] Assumption Extractor
|
v
[3a] Fallacy Detector ----+---- [3b] Bias Auditor
(parallel) | (parallel)
v
[4] Second-Order Thinker
|
v
[5] Epistemic Reasoner
|
v
[6] Tradeoff Articulator
Steps
Step 1 — Question Upgrader
Skill: question-upgrader
Goal: Verify we are solving the right problem before investing in an answer
Input: Problem as currently framed
Output: Upgraded question(s), named framing weakness, hidden assumption in the question, the question being avoided
Skip if: Framing has already been challenged and a better question is agreed on
"Step 1 시작" or "질문부터 업그레이드해줘"
Step 2 — Assumption Extractor
Skill: assumption-extractor
Goal: Surface factual, causal, value, and definitional assumptions the argument rests on
Input: Upgraded framing from Step 1
Output: Assumption map across four categories (load-bearing / significant / peripheral); most dangerous assumption named
Skip if: Assumptions have already been explicitly stress-tested
"Step 2 시작" or "숨겨진 전제 찾아줘"
Step 3a + 3b — Fallacy Detector + Bias Auditor (parallel)
Skills: fallacy-detector and bias-auditor
Goal: Check that the argument structure is sound (3a) and the judgment is free of cognitive distortion (3b)
Input: Core argument and rationale from Step 2
Output 3a: Named fallacies with location in argument; narrative fallacy assessment; worst structural error
Output 3b: Active biases across three layers; primary bias and what it distorts
Skip 3a if: The analysis is a plan with no explicit reasoning chains to evaluate
Skip 3b if: No expressed confidence or judgment calls are present
"Step 3 시작" or "논리 오류랑 편향 같이 봐줘"
Step 4 — Second-Order Thinker
Skill: second-order-thinker
Goal: Map downstream effects across five mechanisms with timing labels
Input: Step 2 assumptions + Step 3 structural and bias findings
Output: Consequence chain; critical second-order effect with time horizon and leading indicator
Skip if: Decision affects only the immediate actor with no systemic or behavioral ripple
"Step 4 시작" or "2차 효과 분석해줘"
Step 5 — Epistemic Reasoner
Skill: epistemic-reasoner
Goal: Separate what is actually known from what is believed; calibrate confidence to evidence
Input: Claims and evidence surfaced across Steps 1–4
Output: Calibrated confidence label per core claim (high / moderate / low / speculation); gap from stated confidence; analogy verdict if applicable
Skip if: All claims are already framed as exploratory with no stated certainty
"Step 5 시작" or "확신 수준 보정해줘"
Step 6 — Tradeoff Articulator
Skill: tradeoff-articulator
Goal: Make the cost structure of the preferred option fully visible — including hidden axes
Input: Full picture from Steps 1–5; option(s) under consideration
Output: Hidden tradeoff axes surfaced; qualitative matrix; concrete opportunity costs; unavoidable tension named
Skip if: No real choice exists — only one viable option remains after prior steps
"Step 6 시작" or "트레이드오프 정리해줘"
State Tracking
Tell me which step you're on and I'll pick up from there:
- "가정은 이미 정리됐어, 2차 효과부터" → Step 4로 직행
- "Step 3부터" → run fallacy-detector and bias-auditor in parallel
What Claude Does / What You Do
| Claude | You |
|---|---|
| Reframes the question; names what makes it weak | Provide the original question or problem framing |
| Scans all four assumption categories; classifies by load | Confirm which assumptions are actually operative |
| Identifies fallacies and biases that are actually present | Decide which findings to act on before proceeding |
| Maps second-order effects across five mechanisms with timing | Validate which downstream effects are plausible in context |
| Calibrates confidence labels to actual evidence | Adjust how you state claims in the final argument |
| Surfaces hidden tradeoff axes; names the unavoidable tension | Make the final call with full cost visibility |
Related Skills
- Individual skills:
cognition:question-upgrader,cognition:assumption-extractor,cognition:fallacy-detector,cognition:bias-auditor,cognition:second-order-thinker,cognition:epistemic-reasoner,cognition:tradeoff-articulator - Before:
think:deep-thinking-workflow(open-ended ideation before stress-testing) - After:
pm:pm-strategy-workflow(when findings inform product strategy)