Délègue les problèmes complexes (math, logique, architecture) à Gemini 3 Deep Think. À utiliser pour réflexion approfondie nécessitant exploration multi-hypothèses.
Delegates complex problems to Gemini for deep, multi-hypothesis reasoning and analysis.
/plugin marketplace add atournayre/claude-marketplace/plugin install review@atournayre-claude-plugin-marketplaceclaude-haiku-4-5-20251001Résoudre des problèmes complexes en déléguant à Gemini qui utilise un mode de réflexion approfondie (thinking step by step, exploration multi-hypothèses).
PROBLEM="$ARGUMENTS"
if [ -z "$PROBLEM" ]; then
echo "Usage: /gemini:think <problem-description>"
echo ""
echo "Exemples:"
echo " /gemini:think 'Comment implémenter un système de saga pour transactions distribuées?'"
echo " /gemini:think 'Quel algorithme de cache optimal pour des lectures fréquentes et écritures rares?'"
exit 1
fi
echo "Problème soumis à Gemini Deep Think:"
echo "$PROBLEM"
echo ""
if ! command -v gemini &> /dev/null; then
echo "Gemini CLI non installé"
exit 1
fi
# Prompt structuré pour maximiser la réflexion
DEEP_THINK_PROMPT="You are an expert problem solver. Analyze this problem thoroughly.
## Problem
$PROBLEM
## Instructions
1. First, understand the problem completely
2. Identify key constraints and requirements
3. Consider multiple approaches
4. Evaluate trade-offs for each approach
5. Recommend the best solution with justification
6. Provide implementation guidance if applicable
Think step by step. Show your reasoning process."
RESPONSE_FILE="/tmp/gemini_think_$(date +%s).txt"
TIMEOUT=300
echo "Réflexion en cours avec Gemini 3 Pro..."
echo ""
# Appel avec timeout
if timeout $TIMEOUT gemini -m gemini-3-pro-preview "$DEEP_THINK_PROMPT" > "$RESPONSE_FILE" 2>&1; then
echo "## Analyse Gemini Deep Think"
echo ""
cat "$RESPONSE_FILE"
else
EXIT_CODE=$?
if [ $EXIT_CODE -eq 124 ]; then
echo "Timeout après ${TIMEOUT}s - problème peut-être trop complexe"
else
echo "Erreur Gemini (code: $EXIT_CODE)"
cat "$RESPONSE_FILE"
fi
fi
# Cleanup
rm -f "$RESPONSE_FILE"
task: "Deep Think Gemini"
status: "terminé"
details:
problem: "$PROBLEM"
model: "gemini-3-pro-preview"
mode: "step-by-step reasoning"
En cas d'échec :
Use this agent when analyzing conversation transcripts to find behaviors worth preventing with hooks. Examples: <example>Context: User is running /hookify command without arguments user: "/hookify" assistant: "I'll analyze the conversation to find behaviors you want to prevent" <commentary>The /hookify command without arguments triggers conversation analysis to find unwanted behaviors.</commentary></example><example>Context: User wants to create hooks from recent frustrations user: "Can you look back at this conversation and help me create hooks for the mistakes you made?" assistant: "I'll use the conversation-analyzer agent to identify the issues and suggest hooks." <commentary>User explicitly asks to analyze conversation for mistakes that should be prevented.</commentary></example>