personal-insightslisted

Analyse les patterns d'utilisation de Claude via l'historique des conversations, la memoire native et les memoires de projets. Pipeline sequentiel en 3 phases : (1) Diagnostic cognitif structure (forces, faiblesses, biais de delegation, risques d'atrophie) ; (2) Hot Seat interactif ou Claude pose des questions provocantes adaptees aux findings via AskUserQuestions ; (3) Plan d'action correctif + ajustements de system prompt enrichis par les reponses du Hot Seat. Declencher des qu'un utilisateur demande "/insights", "analyse mon utilisation", "mes patterns", "diagnostic cognitif", "suis-je trop dependant", "audit de mon usage", "blind spots", "AI dependency check", "personal insights", "bilan", "hot seat", "challenge me", "sois brutal", "trusted advisor", "pose-moi des questions difficiles", ou toute demande d'introspection sur sa facon d'utiliser Claude. Meme un simple "bilan" en contexte peut declencher ce skill.
yannick-bottino/consulting-claude-skills · ★ 0 · AI & Automation · score 72

Install: claude install-skill yannick-bottino/consulting-claude-skills

# Personal Insights ## Purpose This skill produces a structured cognitive diagnostic of how a person uses Claude. It goes beyond usage statistics to analyze the *quality* of the human-AI collaboration: delegation patterns, cognitive autonomy signals, blind spots, and evolution over time. The output is not a report card. It is a mirror designed to help the user become more intentional about what they delegate, what they keep, and how their AI usage shapes their thinking over time. ## Execution Flow The skill runs as a single sequential pipeline in three phases. All three phases execute in the same conversation, without interruption. ### Phase 1: Data Collection + Diagnostic (sections 1-6) Collect data from all memory layers, score conversations, detect patterns, and produce diagnostic sections 1 through 6. Deliver this analysis to the user as a written debrief. Do NOT include the corrective action plan yet. ### Phase 2: Hot Seat (interactive Q&A) Immediately after delivering the diagnostic, transition into an interactive Q&A session. The questions are selected and customized based on the specific findings from Phase 1. Use the `ask_user_input_v0` tool to present questions. **Hot Seat Protocol:** 1. Read `references/hot-seat-questions.md` for the question bank. 2. Based on Phase 1 findings, select 12-15 questions organized in three layers: - **Layer A (Q1-Q5) : Surface patterns.** Probe the most visible findings from the diagnostic. These warm up the user and e