Yuki-CC1006

Socratic Questioning — Development skill for Claude Code

Development community

Never answers until you confirm the real question — Socratic problem-diagnosis skill for Cursor / Claude Code.

How to install Socratic Questioning

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Yuki-CC1006/socratic-questioning and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Socratic Questioning does

Never answers until you confirm the real question — Socratic problem-diagnosis skill for Cursor / Claude Code.

Alternatives in Development

Ars Plan — ARS academic-paper plan mode — Socratic chapter-by-chapter planning 4.5k ★
Domxss — Confirm DOM XSS in a real headless browser — injects canary payloads into params + URL fragment and only repor 4.5k ★
Ralphy — My Ralph Wiggum setup, an autonomous bash script that runs Claude Code, Codex, OpenCode, Cursor agent, Qwen & 2.8k ★

README

Socratic Questioning

**Never answers until you confirm the real question.**

A problem-diagnosis skill for Cursor / Claude Code / Codex-style agents. It turns a messy, emotional, or half-formed prompt into one precise question — then stops and waits for your confirmation before any advice.

This is **not** a teaching tutor that forever refuses to answer (code/docs Socratic style). It is a short diagnostic interview: find the question worth answering.

Install

**Cursor** (personal skill):

mkdir -p ~/.cursor/skills/socratic-questioning
curl -fsSL https://raw.githubusercontent.com/Yuki-CC1006/socratic-questioning/main/SKILL.md \
  -o ~/.cursor/skills/socratic-questioning/SKILL.md

Or clone:

git clone https://github.com/Yuki-CC1006/socratic-questioning.git ~/.cursor/skills/socratic-questioning

**Agent Skills CLI** (if you use `npx skills`):

npx skills add Yuki-CC1006/socratic-questioning

Invoke explicitly, e.g. `/socratic-questioning` or “run Socratic questioning”. Auto-invoke is off (`disable-model-invocation: true`) so it will not interrogate you by accident.

What it does

Steelman your statement before probing
Classify each claim as fact / interpretation / value / goal
Ask one question at a time (max 6); stop when enough
Treat emotion as signal, never as evidence
End with a fixed YAML brief (outcome + reframed question) — no judgment until you say confirm

Outcome types

`outcome`	Meaning
`clarified`	Clearer question found
`aporia`	Built on a false premise or missing fact
`track-change`	Asked A, actually needed B
`already-clear`	Original question was already sharp — do not invent a “better” one

Why this exists

Most “Socratic” skills either teach (never answer) or frame product problems (appetite, segments). Everyday confusion mixes facts with feelings and “should”. Th