feat(backfill): implement backfill tool for generating missing embeddings

internal/config/config.go

@@ -17,6 +17,7 @@ type Config struct {
 	Search        SearchConfig        `yaml:"search"`
 	Logging       LoggingConfig       `yaml:"logging"`
 	Observability ObservabilityConfig `yaml:"observability"`
+	Backfill      BackfillConfig      `yaml:"backfill"`
 }
 
 type ServerConfig struct {
@@ -135,3 +136,12 @@ type ObservabilityConfig struct {
 	MetricsEnabled bool `yaml:"metrics_enabled"`
 	PprofEnabled   bool `yaml:"pprof_enabled"`
 }
+
+type BackfillConfig struct {
+	Enabled         bool          `yaml:"enabled"`
+	RunOnStartup    bool          `yaml:"run_on_startup"`
+	Interval        time.Duration `yaml:"interval"`
+	BatchSize       int           `yaml:"batch_size"`
+	MaxPerRun       int           `yaml:"max_per_run"`
+	IncludeArchived bool          `yaml:"include_archived"`
+}

internal/config/loader.go

@@ -95,6 +95,13 @@ func defaultConfig() Config {
 			Level:  "info",
 			Format: "json",
 		},
+		Backfill: BackfillConfig{
+			Enabled:      false,
+			RunOnStartup: false,
+			Interval:     15 * time.Minute,
+			BatchSize:    20,
+			MaxPerRun:    100,
+		},
 	}
 }
 

internal/config/validate.go

@@ -81,5 +81,14 @@ func (c Config) Validate() error {
 		return fmt.Errorf("invalid config: logging.level is required")
 	}
 
+	if c.Backfill.Enabled {
+		if c.Backfill.BatchSize <= 0 {
+			return fmt.Errorf("invalid config: backfill.batch_size must be greater than zero when backfill is enabled")
+		}
+		if c.Backfill.MaxPerRun < c.Backfill.BatchSize {
+			return fmt.Errorf("invalid config: backfill.max_per_run must be >= backfill.batch_size")
+		}
+	}
+
 	return nil
 }