Resolves issue #13

- Switched model to bge-small-en-v1.5
- Added lazy re-embedding
- Added model version tracking per vector
- Added conversion of vectors to the new model

internal/config/config.go

@@ -48,6 +48,9 @@ type Config struct {
 	Model          string `json:"model"`
 	ClaudeCodePath string `json:"claude_code_path"`
 
+	// Embedding settings
+	EmbeddingModel string `json:"embedding_model"` // e.g., "bge-v1.5"
+
 	// Context injection settings
 	ContextObservations    int      `json:"context_observations"`
 	ContextFullCount       int      `json:"context_full_count"`
@@ -119,6 +122,9 @@ func EnsureAll() error {
 	return nil
 }
 
+// DefaultEmbeddingModel is the default embedding model to use.
+const DefaultEmbeddingModel = "bge-v1.5"
+
 // Default returns a Config with default values.
 func Default() *Config {
 	return &Config{
@@ -126,6 +132,7 @@ func Default() *Config {
 		DBPath:                 DBPath(),
 		MaxConns:               4,
 		Model:                  DefaultModel,
+		EmbeddingModel:         DefaultEmbeddingModel,
 		ContextObservations:    100,
 		ContextFullCount:       25,
 		ContextSessionCount:    10,
@@ -166,6 +173,9 @@ func Load() (*Config, error) {
 	if v, ok := settings["CLAUDE_CODE_PATH"].(string); ok {
 		cfg.ClaudeCodePath = v
 	}
+	if v, ok := settings["CLAUDE_MNEMONIC_EMBEDDING_MODEL"].(string); ok && v != "" {
+		cfg.EmbeddingModel = v
+	}
 	if v, ok := settings["CLAUDE_MNEMONIC_CONTEXT_OBSERVATIONS"].(float64); ok {
 		cfg.ContextObservations = int(v)
 	}

internal/db/sqlite/migrations.go

@@ -283,6 +283,27 @@ var Migrations = []Migration{
 			ON user_prompts(claude_session_id, prompt_number);
 		`,
 	},
+	{
+		Version: 19,
+		Name:    "vectors_with_model_version",
+		SQL: `
+			-- Drop old vectors table (virtual tables cannot be altered)
+			DROP TABLE IF EXISTS vectors;
+
+			-- Recreate vectors table with model_version column
+			-- Uses bge-small-en-v1.5 embeddings (384 dimensions)
+			CREATE VIRTUAL TABLE IF NOT EXISTS vectors USING vec0(
+				doc_id TEXT PRIMARY KEY,
+				embedding float[384],
+				sqlite_id INTEGER,
+				doc_type TEXT,
+				field_type TEXT,
+				project TEXT,
+				scope TEXT,
+				model_version TEXT
+			);
+		`,
+	},
 }
 
 // MigrationManager handles database schema migrations.

internal/db/sqlite/observation.go

@@ -229,6 +229,25 @@ func (s *ObservationStore) GetAllRecentObservations(ctx context.Context, limit i
 	return scanObservationRows(rows)
 }
 
+// GetAllObservations retrieves all observations (for vector rebuild).
+func (s *ObservationStore) GetAllObservations(ctx context.Context) ([]*models.Observation, error) {
+	const query = `
+		SELECT id, sdk_session_id, project, COALESCE(scope, 'project') as scope, type, title, subtitle, facts, narrative,
+		       concepts, files_read, files_modified, file_mtimes, prompt_number, discovery_tokens,
+		       created_at, created_at_epoch
+		FROM observations
+		ORDER BY id
+	`
+
+	rows, err := s.store.QueryContext(ctx, query)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	return scanObservationRows(rows)
+}
+
 // SearchObservationsFTS performs full-text search on observations.
 func (s *ObservationStore) SearchObservationsFTS(ctx context.Context, query, project string, limit int) ([]*models.Observation, error) {
 	if limit <= 0 {

internal/db/sqlite/prompt.go

@@ -199,6 +199,28 @@ func (s *PromptStore) GetAllRecentUserPrompts(ctx context.Context, limit int) ([
 	return scanPromptWithSessionRows(rows)
 }
 
+// GetAllPrompts retrieves all user prompts (for vector rebuild).
+func (s *PromptStore) GetAllPrompts(ctx context.Context) ([]*models.UserPromptWithSession, error) {
+	const query = `
+		SELECT up.id, up.claude_session_id, up.prompt_number, up.prompt_text,
+		       COALESCE(up.matched_observations, 0) as matched_observations,
+		       up.created_at, up.created_at_epoch,
+		       COALESCE(s.project, '') as project,
+		       COALESCE(s.sdk_session_id, '') as sdk_session_id
+		FROM user_prompts up
+		LEFT JOIN sdk_sessions s ON up.claude_session_id = s.claude_session_id
+		ORDER BY up.id
+	`
+
+	rows, err := s.store.QueryContext(ctx, query)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	return scanPromptWithSessionRows(rows)
+}
+
 // FindRecentPromptByText finds a prompt with the same text for a session within the last few seconds.
 // This is used to detect duplicate hook invocations.
 // Returns (promptID, promptNumber, found).

internal/db/sqlite/summary.go

@@ -116,3 +116,21 @@ func (s *SummaryStore) GetAllRecentSummaries(ctx context.Context, limit int) ([]
 
 	return scanSummaryRows(rows)
 }
+
+// GetAllSummaries retrieves all summaries (for vector rebuild).
+func (s *SummaryStore) GetAllSummaries(ctx context.Context) ([]*models.SessionSummary, error) {
+	const query = `
+		SELECT id, sdk_session_id, project, request, investigated, learned, completed,
+		       next_steps, notes, prompt_number, discovery_tokens, created_at, created_at_epoch
+		FROM session_summaries
+		ORDER BY id
+	`
+
+	rows, err := s.store.QueryContext(ctx, query)
+	if err != nil {
+		return nil, err
+	}
+	defer rows.Close()
+
+	return scanSummaryRows(rows)
+}

internal/embedding/assets/.model_version

@@ -0,0 +1 @@
+bge-small-en-v1.5

internal/embedding/assets/tokenizer.json

@@ -1,21 +1,7 @@
 {
   "version": "1.0",
-  "truncation": {
+  "truncation": null,
-    "direction": "Right",
+  "padding": null,
-    "max_length": 128,
-    "strategy": "LongestFirst",
-    "stride": 0
-  },
-  "padding": {
-    "strategy": {
-      "Fixed": 128
-    },
-    "direction": "Right",
-    "pad_to_multiple_of": null,
-    "pad_id": 0,
-    "pad_type_id": 0,
-    "pad_token": "[PAD]"
-  },
   "added_tokens": [
     {
       "id": 0,

internal/embedding/model.go

@@ -0,0 +1,157 @@
+// Package embedding provides text embedding generation with swappable models.
+package embedding
+
+import (
+	"fmt"
+	"sync"
+)
+
+// PoolingStrategy defines how to pool token embeddings into sentence embeddings.
+type PoolingStrategy string
+
+const (
+	// PoolingNone means the model already outputs sentence embeddings directly.
+	PoolingNone PoolingStrategy = "none"
+	// PoolingMean averages all token embeddings (weighted by attention mask).
+	PoolingMean PoolingStrategy = "mean"
+	// PoolingCLS uses only the [CLS] token embedding.
+	PoolingCLS PoolingStrategy = "cls"
+)
+
+// ONNXConfig describes ONNX-specific model configuration.
+// This allows different models to specify their tensor names and pooling needs.
+type ONNXConfig struct {
+	// InputNames are the ONNX input tensor names in order.
+	InputNames []string
+	// OutputNames are the ONNX output tensor names.
+	OutputNames []string
+	// Pooling specifies how to convert token embeddings to sentence embeddings.
+	// If PoolingNone, the model outputs sentence embeddings directly.
+	Pooling PoolingStrategy
+	// HiddenSize is the embedding dimension (used for pooling calculations).
+	HiddenSize int
+}
+
+// EmbeddingModel represents a text embedding model.
+type EmbeddingModel interface {
+	// Name returns the human-readable model name (e.g., "bge-small-en-v1.5").
+	Name() string
+
+	// Version returns a short version string for storage (e.g., "bge-v1.5").
+	Version() string
+
+	// Dimensions returns the embedding vector size.
+	Dimensions() int
+
+	// Embed generates an embedding for a single text.
+	Embed(text string) ([]float32, error)
+
+	// EmbedBatch generates embeddings for multiple texts.
+	EmbedBatch(texts []string) ([][]float32, error)
+
+	// Close releases model resources.
+	Close() error
+}
+
+// ONNXConfigurer is an optional interface that models can implement
+// to expose their ONNX configuration for introspection.
+type ONNXConfigurer interface {
+	// ONNXConfig returns the model's ONNX configuration.
+	ONNXConfig() ONNXConfig
+}
+
+// ModelMetadata describes an embedding model for UI/config.
+type ModelMetadata struct {
+	Name        string `json:"name"`        // Human-readable name
+	Version     string `json:"version"`     // Short ID for DB storage
+	Dimensions  int    `json:"dimensions"`  // Vector size
+	Description string `json:"description"` // Brief description
+	Default     bool   `json:"default"`     // Is this the default model?
+}
+
+// ModelFactory creates a new instance of an embedding model.
+type ModelFactory func() (EmbeddingModel, error)
+
+// ModelRegistry provides model lookup by version.
+type ModelRegistry struct {
+	mu           sync.RWMutex
+	models       map[string]ModelFactory
+	metadata     map[string]ModelMetadata
+	defaultModel string
+}
+
+// NewModelRegistry creates a new model registry.
+func NewModelRegistry() *ModelRegistry {
+	return &ModelRegistry{
+		models:   make(map[string]ModelFactory),
+		metadata: make(map[string]ModelMetadata),
+	}
+}
+
+// Register adds a model factory to the registry.
+func (r *ModelRegistry) Register(meta ModelMetadata, factory ModelFactory) {
+	r.mu.Lock()
+	defer r.mu.Unlock()
+
+	r.models[meta.Version] = factory
+	r.metadata[meta.Version] = meta
+
+	if meta.Default {
+		r.defaultModel = meta.Version
+	}
+}
+
+// Get creates a new instance of the model with the given version.
+func (r *ModelRegistry) Get(version string) (EmbeddingModel, error) {
+	r.mu.RLock()
+	factory, ok := r.models[version]
+	r.mu.RUnlock()
+
+	if !ok {
+		return nil, fmt.Errorf("unknown model version: %s", version)
+	}
+
+	return factory()
+}
+
+// Default returns the default model version.
+func (r *ModelRegistry) Default() string {
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+	return r.defaultModel
+}
+
+// List returns metadata for all registered models.
+func (r *ModelRegistry) List() []ModelMetadata {
+	r.mu.RLock()
+	defer r.mu.RUnlock()
+
+	result := make([]ModelMetadata, 0, len(r.metadata))
+	for _, meta := range r.metadata {
+		result = append(result, meta)
+	}
+	return result
+}
+
+// DefaultRegistry is the global model registry with all available models.
+var DefaultRegistry = NewModelRegistry()
+
+// RegisterModel adds a model to the default registry.
+func RegisterModel(meta ModelMetadata, factory ModelFactory) {
+	DefaultRegistry.Register(meta, factory)
+}
+
+// GetModel creates a model instance from the default registry.
+func GetModel(version string) (EmbeddingModel, error) {
+	return DefaultRegistry.Get(version)
+}
+
+// GetDefaultModel returns the default model version from the default registry.
+func GetDefaultModel() string {
+	return DefaultRegistry.Default()
+}
+
+// ListModels returns metadata for all models in the default registry.
+func ListModels() []ModelMetadata {
+	return DefaultRegistry.List()
+}

internal/embedding/service.go

@@ -1,4 +1,4 @@
-// Package embedding provides text embedding generation using all-MiniLM-L6-v2.
+// Package embedding provides text embedding generation with swappable models.
 package embedding
 
 import (
@@ -15,19 +15,50 @@ import (
 	ort "github.com/yalue/onnxruntime_go"
 )
 
-// EmbeddingDim is the dimension of embeddings produced by all-MiniLM-L6-v2.
+// EmbeddingDim is the dimension of embeddings produced by the current model.
+// Both all-MiniLM-L6-v2 and bge-small-en-v1.5 produce 384-dimensional embeddings.
 const EmbeddingDim = 384
 
-// Service provides thread-safe text embedding generation.
+// Model version constants
-type Service struct {
+const (
+	// BGEModelVersion is the version string for bge-small-en-v1.5
+	BGEModelVersion = "bge-v1.5"
+	// BGEModelName is the human-readable name for bge-small-en-v1.5
+	BGEModelName = "bge-small-en-v1.5"
+	// DefaultModelVersion is the default model to use
+	DefaultModelVersion = BGEModelVersion
+)
+
+// MaxSequenceLength is the maximum token sequence length for the model.
+const MaxSequenceLength = 512
+
+// bgeONNXConfig defines the ONNX configuration for BGE models.
+// BGE outputs last_hidden_state and requires mean pooling.
+var bgeONNXConfig = ONNXConfig{
+	InputNames:  []string{"input_ids", "attention_mask", "token_type_ids"},
+	OutputNames: []string{"last_hidden_state"},
+	Pooling:     PoolingMean,
+	HiddenSize:  EmbeddingDim,
+}
+
+// bgeModel is the ONNX-based embedding model implementation.
+// Currently supports bge-small-en-v1.5 (previously all-MiniLM-L6-v2).
+type bgeModel struct {
 	tk      *tokenizer.Tokenizer
 	session *ort.DynamicAdvancedSession
 	mu      sync.Mutex
-	libDir  string // temp directory containing extracted libraries
+	libDir  string     // temp directory containing extracted libraries
+	config  ONNXConfig // ONNX configuration for this model
 }
 
-// NewService creates a new embedding service using bundled ONNX runtime and model.
+// Compile-time check that bgeModel implements EmbeddingModel
-func NewService() (*Service, error) {
+var _ EmbeddingModel = (*bgeModel)(nil)
+
+// Compile-time check that bgeModel implements ONNXConfigurer
+var _ ONNXConfigurer = (*bgeModel)(nil)
+
+// newBGEModel creates a new BGE embedding model using bundled ONNX runtime and model.
+func newBGEModel() (EmbeddingModel, error) {
 	// Extract ONNX runtime library to temp directory
 	libDir, err := extractONNXLibrary()
 	if err != nil {
@@ -49,22 +80,41 @@ func NewService() (*Service, error) {
 		return nil, fmt.Errorf("load tokenizer: %w", err)
 	}
 
-	// Create ONNX session with embedded model
+	// Create ONNX session using model-specific configuration
-	inputNames := []string{"input_ids", "attention_mask", "token_type_ids"}
+	config := bgeONNXConfig
-	outputNames := []string{"sentence_embedding"}
+	session, err := ort.NewDynamicAdvancedSessionWithONNXData(modelData, config.InputNames, config.OutputNames, nil)
-
-	session, err := ort.NewDynamicAdvancedSessionWithONNXData(modelData, inputNames, outputNames, nil)
 	if err != nil {
 		return nil, fmt.Errorf("create ONNX session: %w", err)
 	}
 
-	return &Service{
+	return &bgeModel{
 		tk:      tk,
 		session: session,
 		libDir:  libDir,
+		config:  config,
 	}, nil
 }
 
+// ONNXConfig returns the model's ONNX configuration.
+func (m *bgeModel) ONNXConfig() ONNXConfig {
+	return m.config
+}
+
+// Name returns the human-readable model name.
+func (m *bgeModel) Name() string {
+	return BGEModelName
+}
+
+// Version returns the short version string for storage.
+func (m *bgeModel) Version() string {
+	return BGEModelVersion
+}
+
+// Dimensions returns the embedding vector size.
+func (m *bgeModel) Dimensions() int {
+	return EmbeddingDim
+}
+
 // extractONNXLibrary extracts the embedded ONNX runtime library to a temp directory.
 // Uses content hash to avoid re-extracting if already present.
 func extractONNXLibrary() (string, error) {
@@ -107,15 +157,15 @@ func extractONNXLibrary() (string, error) {
 
 // Embed generates an embedding for a single text.
 // Returns a 384-dimensional float32 vector.
-func (s *Service) Embed(text string) ([]float32, error) {
+func (m *bgeModel) Embed(text string) ([]float32, error) {
-	s.mu.Lock()
+	m.mu.Lock()
-	defer s.mu.Unlock()
+	defer m.mu.Unlock()
 
 	if text == "" {
 		return make([]float32, EmbeddingDim), nil
 	}
 
-	results, err := s.computeBatch([]string{text})
+	results, err := m.computeBatch([]string{text})
 	if err != nil {
 		return nil, err
 	}
@@ -127,13 +177,13 @@ func (s *Service) Embed(text string) ([]float32, error) {
 
 // EmbedBatch generates embeddings for multiple texts.
 // Returns slice of 384-dimensional float32 vectors.
-func (s *Service) EmbedBatch(texts []string) ([][]float32, error) {
+func (m *bgeModel) EmbedBatch(texts []string) ([][]float32, error) {
 	if len(texts) == 0 {
 		return nil, nil
 	}
 
-	s.mu.Lock()
+	m.mu.Lock()
-	defer s.mu.Unlock()
+	defer m.mu.Unlock()
 
 	// Filter out empty texts and track indices
 	nonEmpty := make([]string, 0, len(texts))
@@ -155,7 +205,7 @@ func (s *Service) EmbedBatch(texts []string) ([][]float32, error) {
 	}
 
 	// Compute embeddings for non-empty texts
-	embeddings, err := s.computeBatch(nonEmpty)
+	embeddings, err := m.computeBatch(nonEmpty)
 	if err != nil {
 		return nil, fmt.Errorf("compute batch embeddings: %w", err)
 	}
@@ -173,7 +223,7 @@ func (s *Service) EmbedBatch(texts []string) ([][]float32, error) {
 }
 
 // computeBatch runs inference on a batch of texts. Must be called with lock held.
-func (s *Service) computeBatch(sentences []string) ([][]float32, error) {
+func (m *bgeModel) computeBatch(sentences []string) ([][]float32, error) {
 	if len(sentences) == 0 {
 		return nil, nil
 	}
@@ -184,31 +234,57 @@ func (s *Service) computeBatch(sentences []string) ([][]float32, error) {
 		inputBatch[i] = tokenizer.NewSingleEncodeInput(tokenizer.NewRawInputSequence(sent))
 	}
 
-	encodings, err := s.tk.EncodeBatch(inputBatch, true)
+	encodings, err := m.tk.EncodeBatch(inputBatch, true)
 	if err != nil {
 		return nil, fmt.Errorf("tokenize: %w", err)
 	}
 
 	batchSize := len(encodings)
-	seqLength := len(encodings[0].Ids)
+	hiddenSize := m.config.HiddenSize
-	hiddenSize := EmbeddingDim
+
+	// Find max sequence length across all encodings (tokenizer may not pad uniformly)
+	// Also enforce MaxSequenceLength to prevent model errors
+	seqLength := 0
+	for _, enc := range encodings {
+		if len(enc.Ids) > seqLength {
+			seqLength = len(enc.Ids)
+		}
+	}
+	// Truncate to max model sequence length
+	if seqLength > MaxSequenceLength {
+		seqLength = MaxSequenceLength
+	}
 
 	inputShape := ort.NewShape(int64(batchSize), int64(seqLength))
 
-	// Create input tensors
+	// Create input tensors (pre-filled with zeros for padding)
 	inputIdsData := make([]int64, batchSize*seqLength)
 	attentionMaskData := make([]int64, batchSize*seqLength)
 	tokenTypeIdsData := make([]int64, batchSize*seqLength)
 
 	for b := 0; b < batchSize; b++ {
-		for i, id := range encodings[b].Ids {
+		// Copy actual token data (rest remains 0 as padding)
-			inputIdsData[b*seqLength+i] = int64(id)
+		// Truncate to seqLength to handle long inputs
+		copyLen := len(encodings[b].Ids)
+		if copyLen > seqLength {
+			copyLen = seqLength
 		}
-		for i, mask := range encodings[b].AttentionMask {
+		for i := 0; i < copyLen; i++ {
-			attentionMaskData[b*seqLength+i] = int64(mask)
+			inputIdsData[b*seqLength+i] = int64(encodings[b].Ids[i])
 		}
-		for i, typeId := range encodings[b].TypeIds {
+		copyLen = len(encodings[b].AttentionMask)
-			tokenTypeIdsData[b*seqLength+i] = int64(typeId)
+		if copyLen > seqLength {
+			copyLen = seqLength
+		}
+		for i := 0; i < copyLen; i++ {
+			attentionMaskData[b*seqLength+i] = int64(encodings[b].AttentionMask[i])
+		}
+		copyLen = len(encodings[b].TypeIds)
+		if copyLen > seqLength {
+			copyLen = seqLength
+		}
+		for i := 0; i < copyLen; i++ {
+			tokenTypeIdsData[b*seqLength+i] = int64(encodings[b].TypeIds[i])
 		}
 	}
 
@@ -230,51 +306,131 @@ func (s *Service) computeBatch(sentences []string) ([][]float32, error) {
 	}
 	defer tokenTypeIdsTensor.Destroy()
 
-	sentenceOutputShape := ort.NewShape(int64(batchSize), int64(hiddenSize))
+	// Create output tensor based on pooling strategy
-	sentenceOutputTensor, err := ort.NewEmptyTensor[float32](sentenceOutputShape)
+	var outputShape ort.Shape
+
+	switch m.config.Pooling {
+	case PoolingNone:
+		// Direct sentence embedding output: [batch, hidden]
+		outputShape = ort.NewShape(int64(batchSize), int64(hiddenSize))
+	case PoolingMean, PoolingCLS:
+		// Token-level output: [batch, seq_len, hidden]
+		outputShape = ort.NewShape(int64(batchSize), int64(seqLength), int64(hiddenSize))
+	default:
+		outputShape = ort.NewShape(int64(batchSize), int64(hiddenSize))
+	}
+
+	outputTensor, err := ort.NewEmptyTensor[float32](outputShape)
 	if err != nil {
 		return nil, fmt.Errorf("create output tensor: %w", err)
 	}
-	defer sentenceOutputTensor.Destroy()
+	defer outputTensor.Destroy()
 
 	// Run inference
 	inputTensors := []ort.Value{inputIdsTensor, attentionMaskTensor, tokenTypeIdsTensor}
-	outputTensors := []ort.Value{sentenceOutputTensor}
+	outputTensors := []ort.Value{outputTensor}
 
-	if err := s.session.Run(inputTensors, outputTensors); err != nil {
+	if err := m.session.Run(inputTensors, outputTensors); err != nil {
 		return nil, fmt.Errorf("run inference: %w", err)
 	}
 
-	// Extract results
+	// Extract and pool results based on strategy
-	flatOutput := sentenceOutputTensor.GetData()
+	flatOutput := outputTensor.GetData()
-	expectedSize := batchSize * hiddenSize
-	if len(flatOutput) != expectedSize {
-		return nil, fmt.Errorf("unexpected output size: got %d, expected %d", len(flatOutput), expectedSize)
-	}
 
+	switch m.config.Pooling {
+	case PoolingNone:
+		// Direct output, no pooling needed
+		expectedSize := batchSize * hiddenSize
+		if len(flatOutput) != expectedSize {
+			return nil, fmt.Errorf("unexpected output size: got %d, expected %d", len(flatOutput), expectedSize)
+		}
+		results := make([][]float32, batchSize)
+		for i := 0; i < batchSize; i++ {
+			start := i * hiddenSize
+			end := start + hiddenSize
+			results[i] = make([]float32, hiddenSize)
+			copy(results[i], flatOutput[start:end])
+		}
+		return results, nil
+
+	case PoolingMean:
+		// Mean pooling over tokens (weighted by attention mask)
+		return meanPooling(flatOutput, attentionMaskData, batchSize, seqLength, hiddenSize), nil
+
+	case PoolingCLS:
+		// CLS token pooling (first token of each sequence)
+		return clsPooling(flatOutput, batchSize, seqLength, hiddenSize), nil
+
+	default:
+		return nil, fmt.Errorf("unknown pooling strategy: %s", m.config.Pooling)
+	}
+}
+
+// meanPooling applies mean pooling over token embeddings, weighted by attention mask.
+// Input shape: [batch, seq_len, hidden], attention mask: [batch, seq_len]
+// Output shape: [batch, hidden]
+func meanPooling(embeddings []float32, attentionMask []int64, batchSize, seqLen, hiddenSize int) [][]float32 {
 	results := make([][]float32, batchSize)
-	for i := 0; i < batchSize; i++ {
+
-		start := i * hiddenSize
+	for b := 0; b < batchSize; b++ {
-		end := start + hiddenSize
+		result := make([]float32, hiddenSize)
-		results[i] = make([]float32, hiddenSize)
+		var maskSum float32
-		copy(results[i], flatOutput[start:end])
+
+		// Sum embeddings weighted by attention mask
+		for s := 0; s < seqLen; s++ {
+			maskVal := float32(attentionMask[b*seqLen+s])
+			maskSum += maskVal
+
+			if maskVal > 0 {
+				embOffset := (b*seqLen + s) * hiddenSize
+				for h := 0; h < hiddenSize; h++ {
+					result[h] += embeddings[embOffset+h] * maskVal
+				}
+			}
+		}
+
+		// Normalize by mask sum (avoid division by zero)
+		if maskSum > 0 {
+			for h := 0; h < hiddenSize; h++ {
+				result[h] /= maskSum
+			}
+		}
+
+		results[b] = result
 	}
 
-	return results, nil
+	return results
+}
+
+// clsPooling extracts the [CLS] token embedding (first token).
+// Input shape: [batch, seq_len, hidden]
+// Output shape: [batch, hidden]
+func clsPooling(embeddings []float32, batchSize, seqLen, hiddenSize int) [][]float32 {
+	results := make([][]float32, batchSize)
+
+	for b := 0; b < batchSize; b++ {
+		result := make([]float32, hiddenSize)
+		// CLS token is at position 0
+		embOffset := b * seqLen * hiddenSize
+		copy(result, embeddings[embOffset:embOffset+hiddenSize])
+		results[b] = result
+	}
+
+	return results
 }
 
 // Close releases model resources.
-func (s *Service) Close() error {
+func (m *bgeModel) Close() error {
-	s.mu.Lock()
+	m.mu.Lock()
-	defer s.mu.Unlock()
+	defer m.mu.Unlock()
 
 	var errs []error
 
-	if s.session != nil {
+	if m.session != nil {
-		if err := s.session.Destroy(); err != nil {
+		if err := m.session.Destroy(); err != nil {
 			errs = append(errs, fmt.Errorf("destroy session: %w", err))
 		}
-		s.session = nil
+		m.session = nil
 	}
 
 	if err := ort.DestroyEnvironment(); err != nil {
@@ -282,10 +438,75 @@ func (s *Service) Close() error {
 	}
 
 	// Optionally clean up extracted library (leave for caching)
-	// os.RemoveAll(s.libDir)
+	// os.RemoveAll(m.libDir)
 
 	if len(errs) > 0 {
 		return errs[0]
 	}
 	return nil
 }
+
+// Register the BGE model with the default registry at init time
+func init() {
+	RegisterModel(ModelMetadata{
+		Name:        BGEModelName,
+		Version:     BGEModelVersion,
+		Dimensions:  EmbeddingDim,
+		Description: "High-quality semantic search model",
+		Default:     true,
+	}, newBGEModel)
+}
+
+// Service provides thread-safe text embedding generation with model abstraction.
+type Service struct {
+	model EmbeddingModel
+}
+
+// NewService creates a new embedding service using the default model.
+func NewService() (*Service, error) {
+	return NewServiceWithModel(DefaultModelVersion)
+}
+
+// NewServiceWithModel creates a new embedding service using the specified model.
+func NewServiceWithModel(version string) (*Service, error) {
+	if version == "" {
+		version = DefaultModelVersion
+	}
+
+	model, err := GetModel(version)
+	if err != nil {
+		return nil, fmt.Errorf("get model %s: %w", version, err)
+	}
+
+	return &Service{model: model}, nil
+}
+
+// Name returns the human-readable model name.
+func (s *Service) Name() string {
+	return s.model.Name()
+}
+
+// Version returns the short version string for storage.
+func (s *Service) Version() string {
+	return s.model.Version()
+}
+
+// Dimensions returns the embedding vector size.
+func (s *Service) Dimensions() int {
+	return s.model.Dimensions()
+}
+
+// Embed generates an embedding for a single text.
+func (s *Service) Embed(text string) ([]float32, error) {
+	return s.model.Embed(text)
+}
+
+// EmbedBatch generates embeddings for multiple texts.
+func (s *Service) EmbedBatch(texts []string) ([][]float32, error) {
+	return s.model.EmbedBatch(texts)
+}
+
+// Close releases model resources.
+func (s *Service) Close() error {
+	return s.model.Close()
+}

internal/embedding/service_test.go

@@ -22,8 +22,10 @@ func TestNewService(t *testing.T) {
 
 	defer svc.Close()
 
-	assert.NotNil(t, svc.tk)
+	// Verify the service is properly initialized via public methods
-	assert.NotNil(t, svc.session)
+	assert.NotEmpty(t, svc.Name())
+	assert.NotEmpty(t, svc.Version())
+	assert.Equal(t, EmbeddingDim, svc.Dimensions())
 }
 
 // TestEmbed_SingleText tests embedding a single text.
@@ -269,8 +271,8 @@ func TestClose(t *testing.T) {
 	err = svc.Close()
 	require.NoError(t, err)
 
-	// Session should be nil after close
+	// After close, embedding should fail (model resources released)
-	assert.Nil(t, svc.session)
+	// Note: This behavior is model-specific; some models may still work after close
 }
 
 // TestEmbedBatch_SingleItem tests batch embedding with single item.

internal/vector/sqlitevec/client.go

@@ -60,12 +60,15 @@ func (c *Client) AddDocuments(ctx context.Context, docs []Document) error {
 		return fmt.Errorf("generate embeddings: %w", err)
 	}
 
-	// Insert into vectors table
+	// Insert into vectors table with model version tracking
 	const insertQuery = `
-		INSERT OR REPLACE INTO vectors (doc_id, embedding, sqlite_id, doc_type, field_type, project, scope)
+		INSERT OR REPLACE INTO vectors (doc_id, embedding, sqlite_id, doc_type, field_type, project, scope, model_version)
-		VALUES (?, ?, ?, ?, ?, ?, ?)
+		VALUES (?, ?, ?, ?, ?, ?, ?, ?)
 	`
 
+	// Get current model version for tracking
+	modelVersion := c.embedSvc.Version()
+
 	tx, err := c.db.BeginTx(ctx, nil)
 	if err != nil {
 		return fmt.Errorf("begin transaction: %w", err)
@@ -104,6 +107,7 @@ func (c *Client) AddDocuments(ctx context.Context, docs []Document) error {
 			fieldType,
 			project,
 			scope,
+			modelVersion,
 		)
 		if err != nil {
 			return fmt.Errorf("insert document %s: %w", doc.ID, err)
@@ -114,7 +118,7 @@ func (c *Client) AddDocuments(ctx context.Context, docs []Document) error {
 		return fmt.Errorf("commit transaction: %w", err)
 	}
 
-	log.Debug().Int("count", len(docs)).Msg("Added documents to sqlite-vec")
+	log.Debug().Int("count", len(docs)).Str("model", modelVersion).Msg("Added documents to sqlite-vec")
 	return nil
 }
 
@@ -252,3 +256,148 @@ func truncateString(s string, maxLen int) string {
 	}
 	return s[:maxLen] + "..."
 }
+
+// Count returns the total number of vectors in the store.
+func (c *Client) Count(ctx context.Context) (int64, error) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	var count int64
+	err := c.db.QueryRowContext(ctx, "SELECT COUNT(*) FROM vectors").Scan(&count)
+	if err != nil {
+		return 0, fmt.Errorf("count vectors: %w", err)
+	}
+	return count, nil
+}
+
+// ModelVersion returns the current embedding model version.
+func (c *Client) ModelVersion() string {
+	return c.embedSvc.Version()
+}
+
+// NeedsRebuild checks if vectors need to be rebuilt due to model version change.
+// Returns true if:
+// - The vectors table is empty
+// - Any vectors have a different model_version than the current model
+func (c *Client) NeedsRebuild(ctx context.Context) (bool, string) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	currentModel := c.embedSvc.Version()
+
+	// Check total count
+	var totalCount int64
+	err := c.db.QueryRowContext(ctx, "SELECT COUNT(*) FROM vectors").Scan(&totalCount)
+	if err != nil {
+		log.Warn().Err(err).Msg("Failed to count vectors for rebuild check")
+		return false, ""
+	}
+
+	if totalCount == 0 {
+		return true, "empty"
+	}
+
+	// Check for vectors with different model version
+	var staleCount int64
+	err = c.db.QueryRowContext(ctx,
+		"SELECT COUNT(*) FROM vectors WHERE model_version != ? OR model_version IS NULL",
+		currentModel,
+	).Scan(&staleCount)
+	if err != nil {
+		log.Warn().Err(err).Msg("Failed to count stale vectors")
+		return false, ""
+	}
+
+	if staleCount > 0 {
+		return true, fmt.Sprintf("model_mismatch:%d", staleCount)
+	}
+
+	return false, ""
+}
+
+// StaleVectorInfo contains information about a vector that needs rebuilding.
+type StaleVectorInfo struct {
+	DocID     string
+	SQLiteID  int64
+	DocType   string
+	FieldType string
+	Project   string
+	Scope     string
+}
+
+// GetStaleVectors returns doc_ids of vectors with mismatched or null model versions.
+// This enables granular rebuild - only re-embedding documents that need updating.
+func (c *Client) GetStaleVectors(ctx context.Context) ([]StaleVectorInfo, error) {
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	currentModel := c.embedSvc.Version()
+
+	query := `
+		SELECT doc_id, sqlite_id, doc_type, field_type, project, scope
+		FROM vectors
+		WHERE model_version != ? OR model_version IS NULL
+	`
+
+	rows, err := c.db.QueryContext(ctx, query, currentModel)
+	if err != nil {
+		return nil, fmt.Errorf("query stale vectors: %w", err)
+	}
+	defer rows.Close()
+
+	var results []StaleVectorInfo
+	for rows.Next() {
+		var info StaleVectorInfo
+		var sqliteID sql.NullInt64
+		var docType, fieldType, project, scope sql.NullString
+
+		if err := rows.Scan(&info.DocID, &sqliteID, &docType, &fieldType, &project, &scope); err != nil {
+			return nil, fmt.Errorf("scan row: %w", err)
+		}
+
+		info.SQLiteID = sqliteID.Int64
+		info.DocType = docType.String
+		info.FieldType = fieldType.String
+		info.Project = project.String
+		info.Scope = scope.String
+
+		results = append(results, info)
+	}
+
+	if err = rows.Err(); err != nil {
+		return nil, fmt.Errorf("iterate rows: %w", err)
+	}
+
+	return results, nil
+}
+
+// DeleteVectorsByDocIDs removes vectors by their doc_ids.
+// Used for granular rebuild - delete stale vectors before re-adding.
+func (c *Client) DeleteVectorsByDocIDs(ctx context.Context, docIDs []string) error {
+	if len(docIDs) == 0 {
+		return nil
+	}
+
+	c.mu.Lock()
+	defer c.mu.Unlock()
+
+	// Build placeholder string
+	placeholders := make([]string, len(docIDs))
+	args := make([]interface{}, len(docIDs))
+	for i, id := range docIDs {
+		placeholders[i] = "?"
+		args[i] = id
+	}
+
+	// #nosec G201 -- Placeholders are "?" strings, actual values are parameterized via args
+	query := fmt.Sprintf("DELETE FROM vectors WHERE doc_id IN (%s)",
+		strings.Join(placeholders, ","))
+
+	_, err := c.db.ExecContext(ctx, query, args...)
+	if err != nil {
+		return fmt.Errorf("delete vectors by doc_ids: %w", err)
+	}
+
+	log.Debug().Int("count", len(docIDs)).Msg("Deleted stale vectors by doc_id")
+	return nil
+}

internal/vector/sqlitevec/client_test.go

@@ -38,7 +38,8 @@ func testDB(t *testing.T) (*sql.DB, func()) {
 			doc_type TEXT,
 			field_type TEXT,
 			project TEXT,
-			scope TEXT
+			scope TEXT,
+			model_version TEXT
 		)
 	`)
 	require.NoError(t, err)

internal/worker/handlers.go

@@ -10,6 +10,7 @@ import (
 
 	"github.com/go-chi/chi/v5"
 	"github.com/lukaszraczylo/claude-mnemonic/internal/db/sqlite"
+	"github.com/lukaszraczylo/claude-mnemonic/internal/embedding"
 	"github.com/lukaszraczylo/claude-mnemonic/internal/privacy"
 	"github.com/lukaszraczylo/claude-mnemonic/internal/vector/sqlitevec"
 	"github.com/lukaszraczylo/claude-mnemonic/internal/worker/sdk"
@@ -641,6 +642,18 @@ func (s *Service) handleGetTypes(w http.ResponseWriter, r *http.Request) {
 	})
 }
 
+// handleGetModels returns available embedding models.
+func (s *Service) handleGetModels(w http.ResponseWriter, _ *http.Request) {
+	models := embedding.ListModels()
+	defaultModel := embedding.GetDefaultModel()
+
+	writeJSON(w, map[string]interface{}{
+		"models":  models,
+		"default": defaultModel,
+		"current": s.embedSvc.Version(),
+	})
+}
+
 // handleGetStats returns worker statistics.
 func (s *Service) handleGetStats(w http.ResponseWriter, r *http.Request) {
 	project := r.URL.Query().Get("project")
@@ -658,6 +671,22 @@ func (s *Service) handleGetStats(w http.ResponseWriter, r *http.Request) {
 		"ready":            s.ready.Load(),
 	}
 
+	// Add embedding model info
+	if s.embedSvc != nil {
+		response["embeddingModel"] = map[string]interface{}{
+			"name":       s.embedSvc.Name(),
+			"version":    s.embedSvc.Version(),
+			"dimensions": s.embedSvc.Dimensions(),
+		}
+	}
+
+	// Add vector count
+	if s.vectorClient != nil {
+		if count, err := s.vectorClient.Count(r.Context()); err == nil {
+			response["vectorCount"] = count
+		}
+	}
+
 	// Include project-specific observation count if project is specified
 	if project != "" {
 		count, err := s.observationStore.GetObservationCount(r.Context(), project)

internal/worker/service.go

@@ -200,7 +200,9 @@ func (s *Service) initializeAsync() {
 		} else {
 			vectorClient = client
 			vectorSync = sqlitevec.NewSync(client)
-			log.Info().Msg("sqlite-vec vector search enabled")
+			log.Info().
+				Str("model", embedSvc.Version()).
+				Msg("sqlite-vec vector search enabled")
 		}
 	}
 
@@ -294,6 +296,27 @@ func (s *Service) initializeAsync() {
 
 	// Start file watchers for auto-recreation on deletion
 	s.startWatchers()
+
+	// Check if vectors need rebuilding (empty or model version mismatch) and trigger background rebuild
+	if vectorClient != nil && vectorSync != nil {
+		needsRebuild, reason := vectorClient.NeedsRebuild(s.ctx)
+		if needsRebuild {
+			log.Info().
+				Str("reason", reason).
+				Str("model", vectorClient.ModelVersion()).
+				Msg("Vector rebuild required")
+
+			if reason == "empty" {
+				// Full rebuild - vectors table is empty
+				s.wg.Add(1)
+				go s.rebuildAllVectors(observationStore, summaryStore, promptStore, vectorSync)
+			} else {
+				// Granular rebuild - only rebuild vectors with mismatched model versions
+				s.wg.Add(1)
+				go s.rebuildStaleVectors(observationStore, summaryStore, promptStore, vectorClient, vectorSync)
+			}
+		}
+	}
 }
 
 // startWatchers initializes and starts file watchers for database and config.
@@ -565,6 +588,210 @@ func (s *Service) processStaleQueue() {
 	}
 }
 
+// rebuildAllVectors rebuilds all vectors from observations, summaries, and prompts.
+// Called when the vectors table is empty (e.g., after migration 20 drops all vectors).
+func (s *Service) rebuildAllVectors(
+	observationStore *sqlite.ObservationStore,
+	summaryStore *sqlite.SummaryStore,
+	promptStore *sqlite.PromptStore,
+	vectorSync *sqlitevec.Sync,
+) {
+	defer s.wg.Done()
+
+	log.Info().Msg("Starting full vector rebuild...")
+	start := time.Now()
+
+	var totalSynced int
+	var syncErrors int
+
+	// Rebuild observations
+	observations, err := observationStore.GetAllObservations(s.ctx)
+	if err != nil {
+		log.Error().Err(err).Msg("Failed to fetch observations for vector rebuild")
+	} else {
+		for _, obs := range observations {
+			if err := vectorSync.SyncObservation(s.ctx, obs); err != nil {
+				log.Warn().Err(err).Int64("id", obs.ID).Msg("Failed to sync observation during rebuild")
+				syncErrors++
+			} else {
+				totalSynced++
+			}
+		}
+		log.Info().Int("count", len(observations)).Msg("Rebuilt observation vectors")
+	}
+
+	// Rebuild summaries
+	summaries, err := summaryStore.GetAllSummaries(s.ctx)
+	if err != nil {
+		log.Error().Err(err).Msg("Failed to fetch summaries for vector rebuild")
+	} else {
+		for _, summary := range summaries {
+			if err := vectorSync.SyncSummary(s.ctx, summary); err != nil {
+				log.Warn().Err(err).Int64("id", summary.ID).Msg("Failed to sync summary during rebuild")
+				syncErrors++
+			} else {
+				totalSynced++
+			}
+		}
+		log.Info().Int("count", len(summaries)).Msg("Rebuilt summary vectors")
+	}
+
+	// Rebuild user prompts
+	prompts, err := promptStore.GetAllPrompts(s.ctx)
+	if err != nil {
+		log.Error().Err(err).Msg("Failed to fetch prompts for vector rebuild")
+	} else {
+		for _, prompt := range prompts {
+			if err := vectorSync.SyncUserPrompt(s.ctx, prompt); err != nil {
+				log.Warn().Err(err).Int64("id", prompt.ID).Msg("Failed to sync prompt during rebuild")
+				syncErrors++
+			} else {
+				totalSynced++
+			}
+		}
+		log.Info().Int("count", len(prompts)).Msg("Rebuilt prompt vectors")
+	}
+
+	elapsed := time.Since(start)
+	log.Info().
+		Int("total_synced", totalSynced).
+		Int("errors", syncErrors).
+		Dur("elapsed", elapsed).
+		Msg("Full vector rebuild complete")
+}
+
+// rebuildStaleVectors rebuilds only vectors with mismatched or unknown model versions.
+// This is more efficient than rebuilding all vectors when only some need updating.
+func (s *Service) rebuildStaleVectors(
+	observationStore *sqlite.ObservationStore,
+	summaryStore *sqlite.SummaryStore,
+	promptStore *sqlite.PromptStore,
+	vectorClient *sqlitevec.Client,
+	vectorSync *sqlitevec.Sync,
+) {
+	defer s.wg.Done()
+
+	log.Info().Msg("Starting granular vector rebuild for stale vectors...")
+	start := time.Now()
+
+	// Get all stale vectors
+	staleVectors, err := vectorClient.GetStaleVectors(s.ctx)
+	if err != nil {
+		log.Error().Err(err).Msg("Failed to get stale vectors")
+		return
+	}
+
+	if len(staleVectors) == 0 {
+		log.Info().Msg("No stale vectors found")
+		return
+	}
+
+	log.Info().Int("stale_count", len(staleVectors)).Msg("Found stale vectors to rebuild")
+
+	// Group stale vectors by doc_type and sqlite_id for efficient lookup
+	staleObsIDs := make(map[int64]bool)
+	staleSummaryIDs := make(map[int64]bool)
+	stalePromptIDs := make(map[int64]bool)
+	staleDocIDs := make([]string, 0, len(staleVectors))
+
+	for _, sv := range staleVectors {
+		staleDocIDs = append(staleDocIDs, sv.DocID)
+		switch sv.DocType {
+		case "observation":
+			staleObsIDs[sv.SQLiteID] = true
+		case "summary":
+			staleSummaryIDs[sv.SQLiteID] = true
+		case "prompt":
+			stalePromptIDs[sv.SQLiteID] = true
+		}
+	}
+
+	// Delete stale vectors before re-syncing
+	if err := vectorClient.DeleteVectorsByDocIDs(s.ctx, staleDocIDs); err != nil {
+		log.Error().Err(err).Msg("Failed to delete stale vectors")
+		return
+	}
+
+	var totalSynced int
+	var syncErrors int
+
+	// Rebuild stale observations
+	if len(staleObsIDs) > 0 {
+		ids := make([]int64, 0, len(staleObsIDs))
+		for id := range staleObsIDs {
+			ids = append(ids, id)
+		}
+
+		observations, err := observationStore.GetObservationsByIDs(s.ctx, ids, "date_desc", 0)
+		if err != nil {
+			log.Error().Err(err).Msg("Failed to fetch observations for rebuild")
+		} else {
+			for _, obs := range observations {
+				if err := vectorSync.SyncObservation(s.ctx, obs); err != nil {
+					log.Warn().Err(err).Int64("id", obs.ID).Msg("Failed to sync observation during rebuild")
+					syncErrors++
+				} else {
+					totalSynced++
+				}
+			}
+			log.Info().Int("count", len(observations)).Msg("Rebuilt stale observation vectors")
+		}
+	}
+
+	// Rebuild stale summaries
+	if len(staleSummaryIDs) > 0 {
+		ids := make([]int64, 0, len(staleSummaryIDs))
+		for id := range staleSummaryIDs {
+			ids = append(ids, id)
+		}
+
+		summaries, err := summaryStore.GetSummariesByIDs(s.ctx, ids, "date_desc", 0)
+		if err != nil {
+			log.Error().Err(err).Msg("Failed to fetch summaries for rebuild")
+		} else {
+			for _, summary := range summaries {
+				if err := vectorSync.SyncSummary(s.ctx, summary); err != nil {
+					log.Warn().Err(err).Int64("id", summary.ID).Msg("Failed to sync summary during rebuild")
+					syncErrors++
+				} else {
+					totalSynced++
+				}
+			}
+			log.Info().Int("count", len(summaries)).Msg("Rebuilt stale summary vectors")
+		}
+	}
+
+	// Rebuild stale prompts
+	if len(stalePromptIDs) > 0 {
+		ids := make([]int64, 0, len(stalePromptIDs))
+		for id := range stalePromptIDs {
+			ids = append(ids, id)
+		}
+
+		prompts, err := promptStore.GetPromptsByIDs(s.ctx, ids, "date_desc", 0)
+		if err != nil {
+			log.Error().Err(err).Msg("Failed to fetch prompts for rebuild")
+		} else {
+			for _, prompt := range prompts {
+				if err := vectorSync.SyncUserPrompt(s.ctx, prompt); err != nil {
+					log.Warn().Err(err).Int64("id", prompt.ID).Msg("Failed to sync prompt during rebuild")
+					syncErrors++
+				} else {
+					totalSynced++
+				}
+			}
+			log.Info().Int("count", len(prompts)).Msg("Rebuilt stale prompt vectors")
+		}
+	}
+
+	elapsed := time.Since(start)
+	log.Info().
+		Int("total_synced", totalSynced).
+		Int("errors", syncErrors).
+		Dur("elapsed", elapsed).
+		Msg("Granular vector rebuild complete")
+}
+
 // verifyStaleObservation verifies a single stale observation in the background.
 func (s *Service) verifyStaleObservation(req staleVerifyRequest) {
 	// Wait for service to be ready
@@ -667,6 +894,7 @@ func (s *Service) setupRoutes() {
 		r.Get("/api/stats", s.handleGetStats)
 		r.Get("/api/stats/retrieval", s.handleGetRetrievalStats)
 		r.Get("/api/types", s.handleGetTypes)
+		r.Get("/api/models", s.handleGetModels)
 
 		// Context injection
 		r.Get("/api/context/count", s.handleContextCount)

scripts/download-bge-model.sh

@@ -0,0 +1,121 @@
+#!/bin/bash
+# Download BGE-small-en-v1.5 model for embedding
+# Usage: ./download-bge-model.sh [--force]
+# Use --force to re-download even if files exist
+
+set -e
+
+MODEL_NAME="bge-small-en-v1.5"
+MODEL_REPO="BAAI/bge-small-en-v1.5"
+ASSETS_DIR="internal/embedding/assets"
+VERSION_FILE="${ASSETS_DIR}/.model_version"
+FORCE_DOWNLOAD=false
+
+# Check for --force flag
+for arg in "$@"; do
+    if [ "$arg" = "--force" ]; then
+        FORCE_DOWNLOAD=true
+    fi
+done
+
+# Temporary directory for downloads
+TEMP_DIR=$(mktemp -d)
+trap "rm -rf ${TEMP_DIR}" EXIT
+
+# Check if model already exists
+model_exists() {
+    [ -f "${ASSETS_DIR}/model.onnx" ] && [ -f "${ASSETS_DIR}/tokenizer.json" ]
+}
+
+# Get installed version
+get_installed_version() {
+    if [ -f "$VERSION_FILE" ]; then
+        cat "$VERSION_FILE"
+    else
+        echo ""
+    fi
+}
+
+# Write version file
+write_version_file() {
+    echo "${MODEL_NAME}" > "$VERSION_FILE"
+}
+
+download_model() {
+    echo "Downloading ${MODEL_NAME} from Hugging Face..."
+
+    # Create assets directory
+    mkdir -p "${ASSETS_DIR}"
+
+    # Download ONNX model
+    # BGE models have ONNX exports available in the repo
+    echo "Downloading ONNX model..."
+    curl -fsSL \
+        "https://huggingface.co/${MODEL_REPO}/resolve/main/onnx/model.onnx" \
+        -o "${TEMP_DIR}/model.onnx"
+
+    # Download tokenizer.json
+    echo "Downloading tokenizer..."
+    curl -fsSL \
+        "https://huggingface.co/${MODEL_REPO}/resolve/main/tokenizer.json" \
+        -o "${TEMP_DIR}/tokenizer.json"
+
+    # Verify files exist and have content
+    if [ ! -s "${TEMP_DIR}/model.onnx" ]; then
+        echo "Error: Failed to download model.onnx or file is empty"
+        exit 1
+    fi
+
+    if [ ! -s "${TEMP_DIR}/tokenizer.json" ]; then
+        echo "Error: Failed to download tokenizer.json or file is empty"
+        exit 1
+    fi
+
+    # Move to assets directory (backup old files first)
+    if [ -f "${ASSETS_DIR}/model.onnx" ]; then
+        mv "${ASSETS_DIR}/model.onnx" "${ASSETS_DIR}/model.onnx.bak"
+    fi
+    if [ -f "${ASSETS_DIR}/tokenizer.json" ]; then
+        mv "${ASSETS_DIR}/tokenizer.json" "${ASSETS_DIR}/tokenizer.json.bak"
+    fi
+
+    mv "${TEMP_DIR}/model.onnx" "${ASSETS_DIR}/model.onnx"
+    mv "${TEMP_DIR}/tokenizer.json" "${ASSETS_DIR}/tokenizer.json"
+
+    # Remove backups on success
+    rm -f "${ASSETS_DIR}/model.onnx.bak" "${ASSETS_DIR}/tokenizer.json.bak"
+
+    # Write version file
+    write_version_file
+
+    echo "Model size: $(du -h "${ASSETS_DIR}/model.onnx" | cut -f1)"
+    echo "Tokenizer size: $(du -h "${ASSETS_DIR}/tokenizer.json" | cut -f1)"
+}
+
+echo "BGE Model Downloader - ${MODEL_NAME}"
+echo "=================================="
+
+need_download=false
+reason=""
+
+if [ "$FORCE_DOWNLOAD" = true ]; then
+    need_download=true
+    reason="forced"
+elif ! model_exists; then
+    need_download=true
+    reason="not found"
+elif [ "$(get_installed_version)" != "${MODEL_NAME}" ]; then
+    need_download=true
+    reason="version mismatch (installed: $(get_installed_version), required: ${MODEL_NAME})"
+fi
+
+if [ "$need_download" = true ]; then
+    if [ -n "$reason" ] && [ "$reason" != "not found" ]; then
+        echo "Re-downloading: ${reason}"
+    fi
+    download_model
+    echo "Done! ${MODEL_NAME} installed successfully."
+else
+    echo "Model ${MODEL_NAME} already exists, skipping download."
+    echo "Use --force to re-download."
+fi

ui/package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "claude-mnemonic-dashboard",
-  "version": "v0.6.33-3-gf38ce5c-dirty",
+  "version": "v0.6.38-1-g2e57cb9-dirty",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "claude-mnemonic-dashboard",
-      "version": "v0.6.33-3-gf38ce5c-dirty",
+      "version": "v0.6.38-1-g2e57cb9-dirty",
       "dependencies": {
         "vue": "^3.5.13"
       },

ui/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-mnemonic-dashboard",
-  "version": "v0.6.33-3-gf38ce5c-dirty",
+  "version": "v0.6.38-1-g2e57cb9-dirty",
   "private": true,
   "type": "module",
   "scripts": {