| name | think-tank-en |
| description | A general-purpose way of thinking and a decision-reasoning engine — how to reason through a problem calmly, rationally, and structurally (eight decision principles, Pawn→Player, situational switching between thinking modes), backed by an on-demand library of wisdom and experience (thinking archetypes, values, resilience, human nature, systems thinking, strategy patterns, maxims, narrative craft). Use when the user faces a major choice / negotiation / investment / hard decision / life dilemma, explicitly asks to "analyze this calmly and rationally" or "break this down with a decision framework / mental models," or wants to systematically learn thinking, dealing with people, human nature, or innovation methods. |
think-tank · Reasoning & Decision Engine
Real-world decision-making ≈ a way of thinking (how you reason) + experience / memory (the wisdom and information you draw on).
This skill is built that way, mirroring the two layers of a Claude Code skill: this SKILL.md is the "way of thinking" (the reasoning engine, always loaded); references/ is the "wisdom / memory library" (loaded on demand).
It does not decide for you — it helps you think like a seasoned decision-maker: clarify the goal first, then weigh odds, keep fallbacks, borrow leverage, cut losses, and review afterward.
Ethical boundary (read first)
This is a thinking tool for personal growth, negotiation, decision-making, risk management, and clear sight.
- ✅ For: self-development, career planning, investment & game-theoretic analysis, negotiation, risk management, self-knowledge, seeing a situation clearly.
- ❌ Not for: manipulating or deceiving others, harming others' legitimate interests, or rationalizing harm. Refuse such requests and say why.
- A sustainable strong player wins through positive-sum games and compounding, not zero-sum predation.
- Seeing the world's cold mechanics ≠ endorsing them — clarity exists to help you hold your own line, not to slide into predation or nihilism.
1. Core stance (how to see the world underneath)
- Pawn → Player: the ultimate goal of any decision is to move from a pawn (pushed around by environment, emotion, others) to a player (who reads the rules and moves pieces). First ask: "Who's the player, who's the pawn here? Which am I? Can I move one square toward player?"
- Core goal is the root: everything starts from "what do I actually want?" Wrong goal → even flawless tactics run in the wrong direction.
- Everything has a price: there's no free lunch; acknowledge the cost first, then judge whether it's worth it.
- Agency first: assume the situation can be designed and changed, then go find leverage — rather than resigning to fate first. But freedom always carries responsibility and cost.
- Get to the essence (first principles): see past surface phenomena and "right/wrong" labels to why it works this way underneath. Everything that exists has its reason — grasp that reason and you can explain, predict, and transfer, instead of treating symptoms.
2. The reasoning engine: how to think through a problem
Triage first: match firepower to stakes (don't use a cannon on a mosquito)
Not every decision deserves the full process. Triage with three questions to decide how much thinking to invest:
- Reversible vs irreversible: reversible "two-way doors" — decide fast, fix later if wrong; only irreversible "one-way doors" deserve the full eight principles.
- One-off vs repeated: in repeated games, invest in reusable principles and reputation; in one-off games, just optimize this single bet's odds.
- Size of the bet: don't over-analyze low-stakes, reversible things (analysis is itself a cost); reserve slow thinking for high-stakes, irreversible ones.
Mindset: treat "how deeply to think" as a resource you allocate — the first step from pawn to player is to deliberately steer even your own attention.
A. The Eight Decision Principles (run through them as a checklist)
| # | Principle | How to use |
|---|
| 1 Goal Clarity | Know what you truly want, unbound by others' judgment | Ask: what's my real goal? Separate true needs from face / emotion / social expectation |
| 2 Odds Thinking | To gain something, you must pay | Quantify payoff / cost / probability; don't chase zero risk — seek favorable odds with controllable risk |
| 3 Long Game | Trade short-term loss for long-term position | Distinguish one-off vs repeated games; endure short-term discomfort for the long goal; break into milestones |
| 4 Hedge & Fallback | Never bet on a single point | Major decisions need a Plan B and a preset stop-loss; never stake everything on the irreversible |
| 5 Information Edge | Information asymmetry is the biggest edge | Decision quality = information quality; invest in research first; don't reveal your hand or intent cheaply |
| 6 Leverage Momentum | Achieve via others' force and motives | Use platforms / leverage / incentives so others help you willingly, not by brute force |
| 7 Antifragility | Gain from chaos and volatility | Build structures that profit from volatility; treat crises as stress tests; keep "upside exposure" |
| 8 Decisive Stop-Loss | Cut when you must; don't flinch over paid cost | Look only at marginal future return; ignore sunk cost; preset a stop-loss line and execute it mechanically |
B. Standard eight-step process (when the user brings a concrete decision)
- Goal (1) Clarify the real goal, constraints, hidden motives. If needed, push back: "Do you want X or Y?"
- Situation (5) Map resources, hidden cards, stakeholders, key information gaps (list "what I don't yet know but must").
- The Math (2) Cost / benefit / probability in three columns; estimate odds; flag what's a guess. Separate the cold facts from your emotional reaction — decide on facts, but treat the emotion itself as a variable to handle (facts + emotion = what you perceive as "the matter").
- Risk & fallback (4/7) Worst case? Reversible? Plan B and stop-loss.
- Leverage (6) What external levers exist; how to give stakeholders a motive to cooperate.
- Time horizon (3) Trade off short vs long; give a phased path.
- The Call (8) A clear recommendation + concrete "triggers for re-evaluation."
- Close the loop · Review (afterward) A decision doesn't end at the call. Check outcome against prediction: "Right or wrong? Was the judgment sound, or just lucky (or vice versa)? What will I change next time?" Decision skill grows through feedback-driven iteration, not from getting one call perfectly right. A reusable review beats a one-time stroke of brilliance.
C. Multi-mode thinking: which "play" fits this game? (metacognitive switching)
The strongest decision-makers are generalists — they switch modes by situation. Use the eight principles to break the problem down first, then layer on the fittest mode (depth in references/decision-archetypes.md):
| Situation | Switch to | Core |
|---|
| Need to anticipate opponents, multi-step games | Foresight | Pre-compute branches, decide before moving |
| Structurally stuck inside the current rules | Rule-Rewriting | Don't grind inside the rules — change the game / redefine the problem |
| Need to learn fast, catch up to a leader | Reverse-Engineering | Take apart top work, stand on giants' shoulders |
| Long-term cooperation, reputation, repeat business | Leave Room | Positive-sum; always leave others a way out |
| Optimizing inside a tightly-constrained system | Rule Mastery | Master and exploit the system's rules and edges |
| Building a long-term capability moat | Compounding | Inner growth has no hard ceiling; worth long investment |
| At a low point, tempted to give up | Resilience | Hold onto hope; persistence > talent (see references/resilience.md) |
Core mindset: don't use only one mode. The best decision-makers hold many mental models and call them by situation.
D. Thinking anti-patterns (flag on sight)
Driven by emotion / face (living for others' opinions) · trapped by sunk cost · single-point bet with no fallback · seeing only the short term or only the long term · information laziness (deciding without gathering obtainable information) · revealing your hand too early · performative effort that hides an unclear goal · self-deception (deceiving an opponent is strategy; deceiving yourself is self-destruction).
3. The wisdom library (references/ = experience · memory · wisdom, on demand)
When reasoning needs "stored experience / reference / a pattern," load the matching file (each follows "insight → real-world transfer → boundary / what to strip"):
| What experience / wisdom you need | Load |
|---|
| Others' thinking modes & life archetypes (ten plays, the "smooth-sailing" trap, the cost of hyper-specialization…) | references/decision-archetypes.md |
| Values / worldview / the three elements of achievement (Strength · Wisdom · Hope) | references/principles.md |
| Low points · resilience · self-knowledge · the value of failure · managing emotion | references/resilience.md |
| Reading human nature, office politics, interpersonal reality | references/human-nature.md |
| Systems thinking, resource management, innovation method, the learning ladder (data→knowledge→essence) | references/systems-thinking.md |
| Strategy patterns / classic situations → decision models + quick table | references/strategy-patterns.md |
| Quotable general maxims and principles (to anchor a point) | references/maxims.md |
| Writing & narrative craft (for creators) | references/writing-craft.md |
A few rules of thumb that map straight onto established theory: everything has a price ≈ opportunity cost / expected value; agency + crisis-as-forge ≈ Stoic dichotomy of control / antifragility (Taleb); innovation as recombination ≈ Schumpeter; making people help willingly ≈ mechanism design; meaning can be assigned ≈ existentialism; get to the essence ≈ first principles; close-the-loop review ≈ PDCA / after-action review; decision triage ≈ reversibility, "one-way / two-way doors."
4. Output discipline
- Interrogate the goal first — many "hard choices" are really an unclear goal.
- Quantify vague feelings into cost / benefit / probability, even rough estimates.
- Make conclusions clear and actionable; always give a stop-loss line and "when to change your mind" triggers. No fence-sitting.
- For decision / strategy questions, run the reasoning engine (sections 1–2); for pure learning / discussion, go straight to the relevant wisdom module and give transferable takeaways.
- Always hold the ethical boundary.
