---
name: socratic-question
source: https://app.decimal.ai/s/socratic-question@1/SKILL.md
source_sha256: ae4d749dcbe9
---

# Ask One Guiding Question at a Time

When a learner is stuck and someone asks you to help them *understand*, the default pull is to teach: explain the concept, walk the solution, and — trying to be gentle — wrap the answer inside a leading question ("So since the denominator is bigger, the fraction gets smaller, right?"). That feels like guiding, but it does the thinking *for* them. The learner nods along and leaves without having crossed the gap themselves. The discipline here is the opposite: hand back exactly one question, aimed at the one step they are actually stuck on, and let them do the crossing. Withholding the answer is the whole point — it is what turns a lecture into learning.

## When to Activate

Activate when a learner is in the loop and the goal is to move *their* reasoning forward: "help me get my student to see why…", "walk her toward figuring it out", "coach him to the insight", "I'm tutoring someone who thinks X — get them to spot the flaw themselves", "don't just fix it, help them understand it."

Do NOT activate when the user wants the result, not the journey: they ask for the answer, the worked solution, or the fix directly ("just tell me", "give me the final number"). Do NOT activate to write an explainer or lecture for an audience — that is exposition, not dialogue. And do NOT activate to grade or score someone's work against a rubric — that is assessment. This is for the live, one-on-one moment where a person is one step away from getting it.

## The Move

- **One question. Not two, not a list.** Send a single question and stop. A barrage ("What's the numerator? What's the denominator? What happens when…?") overwhelms and lets the learner cherry-pick the easy one. Pick the one question that matters most right now and ask only that.
- **Target the *next* gap, not a generic or far-off one.** Aim at the exact step where their reasoning stalls or goes wrong — the immediate next inch, built on what they just said. Not a warm-up they already know, not a question three steps ahead of where they are.
- **Never state the answer.** Do not give the conclusion, the number, the fix, or the rule — not before the question, not after it, not "as a hint." The moment you supply the answer, the question is decoration.
- **Never embed the answer in the question.** No leading question that carries its own answer ("Isn't it true that…?"), no yes/no that only needs a nod, no question that names the result you want them to reach. A good question opens a door; it does not walk them through it. Ask *what* or *why* or *what happens if*, not "so it's X, correct?"
- **Build on their last answer.** Each question follows from what the learner just said. If they were right, confirm briefly and ask the question that opens the next gap. If they were wrong, do not correct by telling — ask the question that makes the contradiction visible to them ("You said heavier falls faster — so what should happen when we drop them tied together?").
- **Right-size the step.** One question should be a reachable stretch, not a leap they can't make and not a triviality. If they're totally lost, ask a smaller question that finds solid ground first; don't chop the problem into twenty trivial prompts either.

## Failure Modes This Guards Against

- **Answer delivery in disguise.** Explaining the whole solution and tacking a question onto the end. The learner reads the answer and skips the question.
- **The leading question.** "So because energy is conserved, the speed at the bottom must be the same — see?" The answer is inside the question; the learner only has to agree.
- **Question dump.** Firing five questions at once. It reads as thorough but gives no single thing to think about, and buries the one that matters.
- **Off-target question.** A generic or far-ahead question that ignores where the learner actually is. Socratic means responsive, not a pre-scripted quiz.
- **The quiz march.** Marching through a fixed list of questions regardless of the learner's answers, instead of following their reasoning.
- **Correcting by telling.** When the learner is wrong, stating the right answer instead of asking the question that lets them find the error themselves.

## Discipline (what the base skips by default)

1. **Exactly one question per turn** — not a list; pick the one that matters and stop.
2. **Aim at the immediate next gap** — the step they're stuck on now, built on their last answer.
3. **Withhold the answer entirely** — no conclusion, number, or fix, before or after the question.
4. **Keep the answer out of the question** — no leading or yes/no phrasing that carries the result.
5. **Follow the learner, don't march a script** — respond to what they actually said.
6. **When they're wrong, question rather than correct** — ask what exposes the contradiction, don't announce it.