M.I.M.I.R - Multi-agent Intelligent Memory & Insight Repository

#!/bin/bash # Ollama Model Setup Script # Intelligently pulls models based on configuration: # - Agent models: Only if Ollama is the defaultProvider in llm-config.json # - Embedding models: Only if MIMIR_EMBEDDINGS_ENABLED=true in .env set -e # Detect Docker Compose command (V1 vs V2) if command -v docker-compose &> /dev/null; then DOCKER_COMPOSE="docker-compose" elif docker compose version &> /dev/null 2>&1; then DOCKER_COMPOSE="docker compose" else DOCKER_COMPOSE="docker compose" # Fallback to V2 fi OLLAMA_CONTAINER="ollama_server" echo "🤖 Setting up Ollama models from config..." echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" # Check if Ollama container is running if ! docker ps | grep -q $OLLAMA_CONTAINER; then echo "❌ Ollama container is not running!" echo " Start it with: $DOCKER_COMPOSE up -d ollama" exit 1 fi echo "✅ Ollama container is running" echo "" # Load environment variables to get embeddings model if [ -f .env ]; then export $(grep -v '^#' .env | grep -v '^$' | xargs) fi # Check if Ollama is the default provider CONFIG_FILE=".mimir/llm-config.json" if [ ! -f "$CONFIG_FILE" ]; then CONFIG_FILE=".mimir/llm-config.example.json" fi DEFAULT_PROVIDER="ollama" # Assume ollama by default if [ -f "$CONFIG_FILE" ]; then # Extract defaultProvider from JSON (using grep/sed for portability) DEFAULT_PROVIDER=$(grep -o '"defaultProvider"[[:space:]]*:[[:space:]]*"[^"]*"' "$CONFIG_FILE" | sed 's/.*"defaultProvider"[[:space:]]*:[[:space:]]*"\([^"]*\)".*/\1/') fi MODELS=() # Only add agent models if Ollama is the default provider if [ "$DEFAULT_PROVIDER" = "ollama" ]; then MODELS+=( "qwen3:8b" # PM/QC agent - agentic capabilities, tool calling (5.2GB) "qwen2.5-coder:1.5b-base" # Worker agent - fast code generation (986MB) ) echo "✅ Ollama is the agent provider - will install agent models" else echo "ℹ️ Ollama is not the default provider ($DEFAULT_PROVIDER) - skipping agent models" fi # Add embedding model if vector embeddings are enabled (independent of agent provider) if [ "${MIMIR_FEATURE_VECTOR_EMBEDDINGS}" = "true" ] || [ "${MIMIR_EMBEDDINGS_ENABLED}" = "true" ]; then EMBEDDING_MODEL="${MIMIR_EMBEDDINGS_MODEL:-mxbai-embed-large}" MODELS+=("$EMBEDDING_MODEL") echo "🧮 Vector embeddings enabled - will also install: $EMBEDDING_MODEL" fi # Exit early if no models to install if [ ${#MODELS[@]} -eq 0 ]; then echo "" echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "ℹ️ No Ollama models needed" echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "" echo "Current configuration:" echo " Default provider: $DEFAULT_PROVIDER" echo " Vector embeddings: ${MIMIR_EMBEDDINGS_ENABLED:-false}" echo "" echo "Ollama is available but not currently configured for:" echo " - Agent execution (using $DEFAULT_PROVIDER instead)" echo " - Vector embeddings (not enabled)" echo "" echo "To use Ollama, update your configuration:" echo " 1. Agent models: Set defaultProvider to 'ollama' in $CONFIG_FILE" echo " 2. Embeddings: Set MIMIR_EMBEDDINGS_ENABLED=true in .env" echo "" exit 0 fi echo "📦 Models to install (from config):" for model in "${MODELS[@]}"; do echo " - $model" done echo "" # Pull each model for model in "${MODELS[@]}"; do echo "📥 Pulling model: $model" # Try normal pull first if docker exec $OLLAMA_CONTAINER ollama pull $model 2>&1 | tee /tmp/ollama_pull_$$.log | grep -q "success"; then echo "✅ Successfully pulled $model" else # Check if it's a TLS certificate error if grep -q "certificate" /tmp/ollama_pull_$$.log 2>/dev/null; then echo "⚠️ TLS certificate error detected. Retrying with insecure mode..." if docker exec -e OLLAMA_INSECURE=true $OLLAMA_CONTAINER ollama pull $model 2>&1 | grep -q "success"; then echo "✅ Successfully pulled $model (used insecure mode)" else echo "❌ Failed to pull $model even with insecure mode" echo " Manual fix: docker exec -e OLLAMA_INSECURE=true $OLLAMA_CONTAINER ollama pull $model" rm -f /tmp/ollama_pull_$$.log exit 1 fi else echo "❌ Failed to pull $model" cat /tmp/ollama_pull_$$.log rm -f /tmp/ollama_pull_$$.log exit 1 fi fi rm -f /tmp/ollama_pull_$$.log echo "" done echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "✨ All models installed successfully!" echo "" echo "Installed models:" docker exec $OLLAMA_CONTAINER ollama list echo "" echo "💾 Storage location: ./data/ollama/models/" echo "📊 Check storage usage: ./scripts/check-storage.sh" echo "" # Show embeddings status if [ "${MIMIR_FEATURE_VECTOR_EMBEDDINGS}" = "true" ] || [ "${MIMIR_EMBEDDINGS_ENABLED}" = "true" ]; then echo "🧮 Vector embeddings configured:" echo " Model: ${MIMIR_EMBEDDINGS_MODEL:-nomic-embed-text}" echo " Dimensions: ${MIMIR_EMBEDDINGS_DIMENSIONS:-768}" echo " Enable file indexing with embeddings: node setup-file-watch.js" echo "" fi echo "━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━" echo "💡 To pull additional models:" echo "" echo " ./scripts/pull-model.sh qwen2.5-coder:7b # Better quality (4.7GB)" echo " ./scripts/pull-model.sh llama3.1:8b # General purpose (4.9GB)" echo " ./scripts/pull-model.sh deepseek-r1:8b # Reasoning (5.2GB)" echo "" echo "🚀 You can now use the agent chain:" echo " npm run chain \"implement authentication\"" echo ""