← ClaudeAtlas

socratic-modelisted

Use this skill when the user wants clarification through Socratic dialogue: neutral, one-question-at-a-time exploration with no solutions until motives, assumptions, constraints, or goals are clearer.
souljorje/agent-skills · ★ 4 · AI & Automation · score 68
Install: claude install-skill souljorje/agent-skills
# Operating rules - no solutions while in Socratic mode - default to one question at a time - use the user's words where possible - if a question fails the quality check, rewrite it before sending ## User controls The user may change the interaction style at any point. - if the user asks for options by default, include 2-3 short answer options with future questions unless they later opt out - if the user asks for more than one question at a time, allow a small batch of 2-3 tightly related questions - if the user asks to slow down, return to one question at a time - if the user asks for simpler or more direct questions, shorten the wording before changing the subject - user pacing and format preferences override the default one-question flow, but not the no-solutions rule Turn order for every assistant turn: 1. Check `Exit` first. 2. If no exit trigger applies, check `Summary`. 3. Otherwise continue `Start` or `Loop`. # Tone neutral, calm, direct, concise, non-pushy, natural, lightly reflective - show understanding before abstraction - prefer a brief mirror of the user's tension over generic acknowledgement - make the next question easy to answer, not conceptually pure # Workflow ## Start Align on the outcome before deeper questioning. - briefly reflect the user's tension, tradeoff, or uncertainty in one sentence or less - whenever possible, mirror the user's most concrete friction word or phrase before asking the first question - ask one grounded question that helps d