Sequential Thinking

sequential-thinking

Sequential thinking is a structured reasoning protocol for problems where jumping to an answer skips the sub-questions that determine whether the answer is right. It works by making the chain of reasoning explicit and checkable — so errors surface at the step level, not buried in a conclusion.

Qué hace esta skill

• Decomposes any complex problem into the sub-questions that must be answered first
• Maps the dependency order between sub-questions (which must be resolved before which)
• Resolves each sub-question explicitly: question stated → evidence reviewed → conclusion drawn
• Synthesizes sub-conclusions into a final answer, clearly separated from the reasoning chain
• Runs a sanity check: "Does this conclusion hold if assumption X is wrong?"

Cuándo usarla

• Debugging a logic error where the symptoms don't point obviously to the cause
• Making a system design or architecture decision with multiple interdependent constraints
• Strategy or pricing decisions where the right answer depends on several prior judgments
• Any problem where your first instinct might be anchoring on the wrong frame
• When someone gives you a recommendation and you want to stress-test it

Proceso

1. State the problem clearly: One sentence — what decision needs to be made or what question answered?

2. Problem decomposition:

   "To answer this, I first need to know: [A], [B], [C]" List every sub-question. Don't skip ones that feel obvious — they contain hidden assumptions.

3. Dependency map: Which sub-questions depend on others being answered first?

   A → B → D
        ↘ C → D
   

   Resolve in dependency order, not in the order you thought of them.

4. Resolve each sub-question explicitly:

   Sub-question: <state it exactly>
   Evidence: <what we know, what we can observe, what's uncertain>
   Conclusion: <single declarative statement>
   Confidence: High / Medium / Low
   

5. Synthesis: Combine sub-conclusions into the final answer. State it in one paragraph first, then elaborate.

6. Sanity check: Pick the 2 most load-bearing assumptions. Ask: "If [assumption] is wrong, does the conclusion reverse?"

   • If yes → flag as a risk, get evidence on that assumption
   • If no → conclusion is robust, proceed

Anti-patterns this skill catches

• Jumping to solution before the problem is understood
• Treating a symptom as a root cause (the error is at a different layer)
• Making multiple assumptions at once without tracking them
• Anchoring to the first idea and reasoning backward to justify it
• Confusing "I don't know" with "probably fine"

Output structure

PROBLEM: <one sentence>

SUB-QUESTIONS:
1. [A] — depends on: nothing
2. [B] — depends on: [A]
3. [C] — depends on: [A], [B]

RESOLUTION:
[A]: <evidence> → <conclusion> (Confidence: High)
[B]: <evidence> → <conclusion> (Confidence: Medium)
[C]: <evidence> → <conclusion> (Confidence: High)

FINAL ANSWER:
<synthesis paragraph>

SANITY CHECK:
If [assumption X] is wrong → [impact on conclusion]
If [assumption Y] is wrong → [impact on conclusion]

Ejemplo de prompt

Estoy pensando en cobrar $99/mes en lugar de $49/mes. No sé si aumentar el precio va a matarme las conversiones o si en realidad me ayuda. Ayúdame a razonar esto paso a paso antes de decidir.

Relación con otras skills