feat: harden plugin system wiring and add developer diagnostics

Fix critical wiring bugs that prevented plugin-provided implementations
from being used at runtime. All CRUD operations, fork/checkout/clear,
batch embedding, neural APIs, and VFS path resolution now properly
dispatch through the plugin registry.

Changes:
- Wire graphIndex to storage for getVerbsBySource() fast path
- Replace instanceof checks with duck-typing (indexIsTypeAware flag)
  so plugin HNSW indexes work in add/update/delete/search
- Add createIndex() shared helper for plugin HNSW factory
- Fix fork/checkout/clear to use plugin factories for metadataIndex,
  graphIndex, and HNSW instead of hardcoding JS constructors
- Add three-tier embedBatch priority: embedBatch > embeddings > WASM
- Skip WASM warmup/eagerEmbeddings when plugin provides embeddings
- Fix PathResolver metadataIndex access (was looking on storage)
- Use global UnifiedCache in SemanticPathResolver
- Wire plugin distance function through neural APIs
- Add diagnostics() method and CLI command for provider inspection
- Add requireProviders() for production fail-fast assertions
- Add init-time provider summary log
- Add plugin developer documentation (docs/PLUGINS.md)
- Export DiagnosticsResult type

docs/PLUGINS.md

@@ -0,0 +1,386 @@
+# Plugin Development Guide
+
+Brainy has a plugin system that allows third-party packages to replace internal subsystems with custom implementations. This is how `@soulcraft/cortex` provides native Rust acceleration, and it's the same system available to any developer.
+
+## Architecture Overview
+
+Brainy's plugin system uses **named providers** — string keys mapped to implementations. During `init()`, brainy:
+
+1. Auto-detects installed plugin packages via dynamic `import()`
+2. Activates each plugin, passing a `BrainyPluginContext`
+3. The plugin calls `context.registerProvider(key, implementation)` for each subsystem it provides
+4. Brainy checks each provider key and wires the implementation into its internal pipeline
+
+If no plugin provides a given key, brainy uses its built-in JavaScript implementation. This means brainy works perfectly standalone — plugins only enhance performance or add capabilities.
+
+## Creating a Plugin
+
+### 1. Implement the `BrainyPlugin` interface
+
+```typescript
+import type { BrainyPlugin, BrainyPluginContext } from '@soulcraft/brainy/plugin'
+
+const myPlugin: BrainyPlugin = {
+  name: 'my-brainy-plugin',  // Must be unique (typically your npm package name)
+
+  async activate(context: BrainyPluginContext): Promise<boolean> {
+    // Register your providers here
+    context.registerProvider('distance', myFastDistanceFunction)
+
+    // Return true if activation succeeded, false to skip
+    return true
+  },
+
+  async deactivate(): Promise<void> {
+    // Optional cleanup when brainy.close() is called
+  }
+}
+
+export default myPlugin
+```
+
+### 2. Package exports
+
+Your package must export the plugin as the default export so brainy's auto-detection works:
+
+```typescript
+// index.ts
+export { default } from './plugin.js'
+```
+
+### 3. Registration
+
+**Auto-detection:** Brainy auto-detects plugins by package name. Pass your package name in the brainy config:
+
+```typescript
+const brain = new Brainy({
+  plugins: ['my-brainy-plugin']
+})
+await brain.init()
+```
+
+**Manual registration:** For plugins not installed as npm packages:
+
+```typescript
+import { Brainy } from '@soulcraft/brainy'
+import myPlugin from './my-plugin.js'
+
+const brain = new Brainy()
+brain.pluginRegistry.register(myPlugin)
+await brain.init()
+```
+
+## Provider Keys Reference
+
+Each key has a specific expected signature. Brainy checks for these during `init()` and wires them into the appropriate code paths.
+
+### Core Providers
+
+#### `distance`
+**Type:** `(a: number[], b: number[]) => number`
+
+Replaces the default cosine distance function used in HNSW search and neural APIs. This is the highest-impact single provider — it's called for every vector comparison.
+
+```typescript
+context.registerProvider('distance', (a: number[], b: number[]): number => {
+  // Your SIMD-accelerated or GPU distance calculation
+  return myFastCosineDistance(a, b)
+})
+```
+
+#### `embeddings`
+**Type:** `(text: string | string[]) => Promise<number[] | number[][]>`
+
+Replaces the built-in WASM embedding engine. Called for every `brain.add()`, `brain.update()`, and `brain.find()` operation that involves text.
+
+```typescript
+context.registerProvider('embeddings', async (text: string | string[]) => {
+  if (Array.isArray(text)) {
+    return myEngine.embedBatch(text)
+  }
+  return myEngine.embed(text)
+})
+```
+
+#### `embedBatch`
+**Type:** `(texts: string[]) => Promise<number[][]>`
+
+Dedicated batch embedding provider. When registered, brainy uses this for bulk operations (import, reindex, batch add) instead of calling the `embeddings` provider N times. This enables true single-forward-pass batch processing.
+
+Priority order for batch operations:
+1. `embedBatch` provider (single forward pass — fastest)
+2. `embeddings` provider with `Promise.all()` (N individual calls)
+3. Built-in WASM batch API (fallback)
+
+```typescript
+context.registerProvider('embedBatch', async (texts: string[]) => {
+  // Process all texts in a single forward pass
+  return myEngine.batchEmbed(texts)
+})
+```
+
+### Index Providers
+
+#### `hnsw`
+**Type:** `(config: object, distanceFunction: Function, options: object) => HNSWIndex-compatible`
+
+Factory function that creates an HNSW index instance. The returned object must implement the `HNSWIndex` public API:
+
+- `addItem(item: { id: string, vector: number[] }): Promise<string>`
+- `search(queryVector: number[], k: number, filter?, options?): Promise<Array<[string, number]>>`
+- `removeItem(id: string): Promise<boolean>`
+- `size(): number`
+- `clear(): void`
+- `flush(): Promise<number>`
+- `rebuild(options?): Promise<void>`
+- `getDirtyNodeCount(): number`
+- `getPersistMode(): 'immediate' | 'deferred'`
+- `getEntryPointId(): string | null`
+- `getMaxLevel(): number`
+- `getDimension(): number | null`
+- `getConfig(): object`
+- `getDistanceFunction(): Function`
+- `enableCOW(parent): void`
+- `setUseParallelization(boolean): void`
+
+For type-aware indexes (separate graph per noun type), also implement:
+- `getIndexForType(type: string): HNSWIndex` (duck-typed detection)
+- `search(queryVector, k, type?, filter?, options?): Promise<Array<[string, number]>>`
+
+```typescript
+context.registerProvider('hnsw', (config, distanceFn, options) => {
+  return new MyNativeHNSWIndex(config, distanceFn, options)
+})
+```
+
+#### `metadataIndex`
+**Type:** `(storage: StorageAdapter) => MetadataIndexManager-compatible`
+
+Factory function that creates a metadata index. The returned object must implement the `MetadataIndexManager` interface including `init()`, `addEntity()`, `removeEntity()`, `query()`, `flush()`, `clear()`, etc.
+
+```typescript
+context.registerProvider('metadataIndex', (storage) => {
+  return new MyNativeMetadataIndex(storage)
+})
+```
+
+#### `graphIndex`
+**Type:** `(storage: StorageAdapter) => GraphAdjacencyIndex-compatible`
+
+Factory function that creates a graph adjacency index for relationship tracking (verbs/triples). Must implement the `GraphAdjacencyIndex` interface including `addVerb()`, `getVerbsBySource()`, `getVerbsByTarget()`, `flush()`, etc.
+
+```typescript
+context.registerProvider('graphIndex', (storage) => {
+  return new MyNativeGraphIndex(storage)
+})
+```
+
+### Utility Providers
+
+#### `cache`
+**Type:** `UnifiedCache`
+
+Replaces the global `UnifiedCache` singleton used for VFS path resolution, semantic caching, and HNSW vector caching. Must implement the `UnifiedCache` interface (available from `@soulcraft/brainy/internals`).
+
+```typescript
+import type { UnifiedCache } from '@soulcraft/brainy/internals'
+
+context.registerProvider('cache', myNativeCache)
+```
+
+#### `entityIdMapper`
+**Type:** `(storage: StorageAdapter) => EntityIdMapper-compatible`
+
+Factory for bidirectional UUID ↔ integer mapping used by roaring bitmaps. Must implement `getOrAssign()`, `getUuid()`, `getInt()`, `has()`, `remove()`, `flush()`, `clear()`.
+
+#### `roaring`
+**Type:** `RoaringBitmap32 class`
+
+Replacement for the roaring bitmap implementation. Used internally by the metadata index for set operations. Must be API-compatible with `roaring-wasm`.
+
+#### `msgpack`
+**Type:** `{ encode: (data: any) => Buffer, decode: (buffer: Buffer) => any }`
+
+Native msgpack encode/decode for SSTable serialization.
+
+## Storage Adapter Plugins
+
+Plugins can register custom storage backends that users reference by name.
+
+### Implementing a Storage Adapter
+
+```typescript
+import type { StorageAdapterFactory } from '@soulcraft/brainy/plugin'
+import type { StorageAdapter } from '@soulcraft/brainy'
+
+class MyStorageAdapter implements StorageAdapter {
+  async init(): Promise<void> { /* ... */ }
+  async saveNoun(noun: HNSWNoun): Promise<void> { /* ... */ }
+  async getNoun(id: string): Promise<HNSWNounWithMetadata | null> { /* ... */ }
+  async deleteNoun(id: string): Promise<void> { /* ... */ }
+  // ... implement all StorageAdapter methods
+}
+```
+
+### Registering a Storage Adapter
+
+```typescript
+context.registerProvider('storage:my-backend', {
+  name: 'my-backend',
+  create: (config: Record<string, unknown>) => {
+    return new MyStorageAdapter(config)
+  }
+} satisfies StorageAdapterFactory)
+```
+
+Users can then use your storage:
+
+```typescript
+const brain = new Brainy({ storage: 'my-backend', myBackendOption: 'value' })
+```
+
+## Import Paths
+
+Brainy provides three entry points for plugin developers:
+
+| Import Path | Contents | Stability |
+|-------------|----------|-----------|
+| `@soulcraft/brainy` | Public API, types, StorageAdapter | Stable (semver) |
+| `@soulcraft/brainy/plugin` | BrainyPlugin, BrainyPluginContext, StorageAdapterFactory | Stable (semver) |
+| `@soulcraft/brainy/internals` | UnifiedCache, EntityIdMapper, logger utilities | Internal (may change between minor versions) |
+
+## Diagnostics
+
+Brainy provides a `diagnostics()` method to verify plugin wiring:
+
+```typescript
+const brain = new Brainy()
+await brain.init()
+
+const diag = brain.diagnostics()
+console.log(diag)
+// {
+//   version: '7.14.0',
+//   plugins: { active: ['my-plugin'], count: 1 },
+//   providers: {
+//     metadataIndex: { source: 'default' },
+//     graphIndex: { source: 'default' },
+//     embeddings: { source: 'plugin' },
+//     embedBatch: { source: 'plugin' },
+//     distance: { source: 'plugin' },
+//     hnsw: { source: 'default' },
+//     ...
+//   },
+//   indexes: {
+//     hnsw: { size: 0, type: 'TypeAwareHNSWIndex' },
+//     metadata: { type: 'MetadataIndexManager', initialized: true },
+//     graph: { type: 'GraphAdjacencyIndex', initialized: true, wiredToStorage: true }
+//   }
+// }
+```
+
+The CLI also supports diagnostics:
+
+```bash
+brainy diagnostics
+```
+
+### Init-Time Summary
+
+When a plugin is active, brainy automatically logs a provider summary after `init()`:
+
+```
+[brainy] Plugin activated: @soulcraft/cortex
+[brainy] Providers: 8/10 native (@soulcraft/cortex) | default: hnsw, cache
+```
+
+This tells you at a glance how many subsystems are accelerated and which ones are falling back to JavaScript. The log respects `config.silent`.
+
+### Fail-Fast for Production
+
+Use `requireProviders()` after `init()` to guarantee specific providers are plugin-supplied. This prevents silent fallback to JavaScript in deployments where you expect native acceleration:
+
+```typescript
+const brain = new Brainy()
+await brain.init()
+
+// Throws immediately if any of these are using JS fallback
+brain.requireProviders(['distance', 'embeddings', 'metadataIndex', 'graphIndex'])
+```
+
+If a required provider is missing, the error message tells you exactly what's wrong:
+
+```
+[brainy] Required providers using JS fallback: graphIndex.
+Active plugins: @soulcraft/cortex.
+These providers must be supplied by a plugin for this deployment.
+Check plugin installation, license, and native module availability.
+```
+
+This is the recommended pattern for production deployments with paid plugins — fail at startup rather than silently degrading performance.
+
+## Complete Example: Distance Acceleration Plugin
+
+A minimal but useful plugin that provides SIMD-accelerated distance calculations:
+
+```typescript
+// simd-distance-plugin/src/plugin.ts
+import type { BrainyPlugin, BrainyPluginContext } from '@soulcraft/brainy/plugin'
+
+// Hypothetical native module
+import { simdCosineDistance } from './native.js'
+
+const simdDistancePlugin: BrainyPlugin = {
+  name: 'brainy-simd-distance',
+
+  async activate(context: BrainyPluginContext): Promise<boolean> {
+    // Check if SIMD is available on this platform
+    if (!checkSimdSupport()) {
+      console.log('[simd-distance] SIMD not available, skipping')
+      return false  // Don't activate — brainy uses JS fallback
+    }
+
+    context.registerProvider('distance', simdCosineDistance)
+    return true
+  }
+}
+
+export default simdDistancePlugin
+```
+
+```json
+// simd-distance-plugin/package.json
+{
+  "name": "brainy-simd-distance",
+  "main": "./dist/plugin.js",
+  "types": "./dist/plugin.d.ts",
+  "peerDependencies": {
+    "@soulcraft/brainy": ">=7.0.0"
+  }
+}
+```
+
+Usage:
+
+```typescript
+import { Brainy } from '@soulcraft/brainy'
+
+const brain = new Brainy({ plugins: ['brainy-simd-distance'] })
+await brain.init()
+
+// Verify it's active
+const diag = brain.diagnostics()
+console.log(diag.providers.distance) // { source: 'plugin' }
+```
+
+## Design Principles
+
+1. **Brainy works perfectly without plugins.** Every provider has a JavaScript fallback. Plugins only improve performance or add capabilities.
+
+2. **Provider keys are string-based.** The plugin system is not coupled to any specific plugin. Any package can register any provider.
+
+3. **Clean separation.** Plugins access brainy through the documented `BrainyPluginContext` interface. No direct access to internal classes is needed.
+
+4. **Fail-safe activation.** If a plugin throws during `activate()`, brainy logs a warning and continues with defaults. A broken plugin never prevents brainy from working.
+
+5. **Lifecycle management.** `deactivate()` is called during `brainy.close()` for resource cleanup. Native resources, connections, and file handles should be released here.

src/brainy.ts

@@ -109,6 +109,20 @@ const STOPWORDS = new Set([
   'what', 'which', 'who', 'whom', 'whose', 'am'
 ])
 
+/**
+ * Result type for brain.diagnostics()
+ */
+export interface DiagnosticsResult {
+  version: string
+  plugins: { active: string[], count: number }
+  providers: Record<string, { source: 'plugin' | 'default' }>
+  indexes: {
+    hnsw: { size: number, type: string }
+    metadata: { type: string, initialized: boolean }
+    graph: { type: string, initialized: boolean, wiredToStorage: boolean }
+  }
+}
+
 /**
  * The main Brainy class - Clean, Beautiful, Powerful
  * REAL IMPLEMENTATION - No stubs, no mocks
@@ -122,6 +136,7 @@ export class Brainy<T = any> implements BrainyInterface<T> {
 
   // Core components
   private index!: HNSWIndex | TypeAwareHNSWIndex
+  private indexIsTypeAware = false  // true when index supports addItem(item, type) / search(v, k, type)
   private storage!: BaseStorage
   private metadataIndex!: MetadataIndexManager
   private graphIndex!: GraphAdjacencyIndex
@@ -292,18 +307,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
         this.distance = nativeDistance
       }
 
-      // Provider: HNSW index factory
+      // Provider: HNSW index factory (plugin or JS fallback)
-      const hnswFactory = this.pluginRegistry.getProvider<(config: any, distance: DistanceFunction, options: any) => any>('hnsw')
+      this.index = this.createIndex()
-      if (hnswFactory) {
+      this.indexIsTypeAware = this.index instanceof TypeAwareHNSWIndex
-        const persistMode = this.resolveHNSWPersistMode()
+        || typeof (this.index as any).getIndexForType === 'function'
-        this.index = hnswFactory(
-          { ...this.config.index, distanceFunction: this.distance },
-          this.distance,
-          { storage: this.storage, persistMode }
-        )
-      } else {
-        this.index = this.setupIndex()
-      }
 
       // Provider: metadata index factory
       const metadataFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('metadataIndex')
@@ -315,6 +322,7 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       const graphFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('graphIndex')
       if (graphFactory) {
         this.graphIndex = graphFactory(this.storage)
+        this.storage.setGraphIndex(this.graphIndex)
         await this.metadataIndex.init()
       } else {
         const [, graphIndex] = await Promise.all([
@@ -350,6 +358,23 @@ export class Brainy<T = any> implements BrainyInterface<T> {
         })
       }
 
+      // Log provider summary after all wiring is complete
+      // Shows developers exactly what's native vs falling back to JS
+      if (this.pluginRegistry.hasActivePlugins() && !this.config.silent) {
+        const wellKnownKeys = [
+          'metadataIndex', 'graphIndex', 'entityIdMapper', 'cache',
+          'hnsw', 'roaring', 'embeddings', 'embedBatch', 'distance', 'msgpack'
+        ]
+        const native = wellKnownKeys.filter(k => this.pluginRegistry.hasProvider(k))
+        const fallback = wellKnownKeys.filter(k => !this.pluginRegistry.hasProvider(k))
+        const plugins = this.pluginRegistry.getActivePlugins().join(', ')
+        if (fallback.length === 0) {
+          console.log(`[brainy] Providers: ${native.length}/${wellKnownKeys.length} native (${plugins})`)
+        } else {
+          console.log(`[brainy] Providers: ${native.length}/${wellKnownKeys.length} native (${plugins}) | default: ${fallback.join(', ')}`)
+        }
+      }
+
       // Mark as initialized BEFORE VFS init
       // VFS.init() needs brain to be marked initialized to call brain methods
       this.initialized = true
@@ -365,10 +390,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       // instead of lazily on first embed() call. This moves the 90-140 second
       // WASM compilation to container startup rather than first request.
       // Recommended for: Cloud Run, Lambda, Fargate, Kubernetes
-      if (this.config.eagerEmbeddings) {
+      if (this.config.eagerEmbeddings && !this.pluginRegistry.hasProvider('embeddings')) {
-        console.log('🚀 Eager embedding initialization enabled...')
+        console.log('Eager embedding initialization enabled...')
         await embeddingManager.init()
-        console.log('✅ Embedding engine ready')
+        console.log('Embedding engine ready')
       }
 
       // Integration Hub initialization
@@ -730,10 +755,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
         )
 
         // Operation 3: Add to HNSW index (after entity saved)
-        if (this.index instanceof TypeAwareHNSWIndex) {
+        if (this.indexIsTypeAware) {
           tx.addOperation(
             new AddToTypeAwareHNSWOperation(
-              this.index,
+              this.index as any,
               id,
               vector,
               params.type as any
@@ -741,7 +766,7 @@ export class Brainy<T = any> implements BrainyInterface<T> {
           )
         } else {
           tx.addOperation(
-            new AddToHNSWOperation(this.index, id, vector)
+            new AddToHNSWOperation(this.index as any, id, vector)
           )
         }
 
@@ -1157,10 +1182,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
 
         // Operation 3-4: Update HNSW index (remove and re-add if reindexing needed)
         if (needsReindexing) {
-          if (this.index instanceof TypeAwareHNSWIndex) {
+          if (this.indexIsTypeAware) {
             tx.addOperation(
               new RemoveFromTypeAwareHNSWOperation(
-                this.index,
+                this.index as any,
                 params.id,
                 existing.vector,
                 existing.type as any
@@ -1168,7 +1193,7 @@ export class Brainy<T = any> implements BrainyInterface<T> {
             )
             tx.addOperation(
               new AddToTypeAwareHNSWOperation(
-                this.index,
+                this.index as any,
                 params.id,
                 vector,
                 newType as any
@@ -1176,10 +1201,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
             )
           } else {
             tx.addOperation(
-              new RemoveFromHNSWOperation(this.index, params.id, existing.vector)
+              new RemoveFromHNSWOperation(this.index as any, params.id, existing.vector)
             )
             tx.addOperation(
-              new AddToHNSWOperation(this.index, params.id, vector)
+              new AddToHNSWOperation(this.index as any, params.id, vector)
             )
           }
         }
@@ -1240,18 +1265,18 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       await this.transactionManager.executeTransaction(async (tx) => {
         // Operation 1: Remove from vector index
         if (noun && metadata) {
-          if (this.index instanceof TypeAwareHNSWIndex && metadata.noun) {
+          if (this.indexIsTypeAware && metadata.noun) {
             tx.addOperation(
               new RemoveFromTypeAwareHNSWOperation(
-                this.index,
+                this.index as any,
                 id,
                 noun.vector,
                 metadata.noun as any
               )
             )
-          } else if (this.index instanceof HNSWIndex) {
+          } else {
             tx.addOperation(
-              new RemoveFromHNSWOperation(this.index, id, noun.vector)
+              new RemoveFromHNSWOperation(this.index as any, id, noun.vector)
             )
           }
         }
@@ -2505,18 +2530,18 @@ export class Brainy<T = any> implements BrainyInterface<T> {
 
               // Add delete operations to transaction
               if (noun && metadata) {
-                if (this.index instanceof TypeAwareHNSWIndex && metadata.noun) {
+                if (this.indexIsTypeAware && metadata.noun) {
                   tx.addOperation(
                     new RemoveFromTypeAwareHNSWOperation(
-                      this.index,
+                      this.index as any,
                       id,
                       noun.vector,
                       metadata.noun as any
                     )
                   )
-                } else if (this.index instanceof HNSWIndex) {
+                } else {
                   tx.addOperation(
-                    new RemoveFromHNSWOperation(this.index, id, noun.vector)
+                    new RemoveFromHNSWOperation(this.index as any, id, noun.vector)
                   )
                 }
               }
@@ -2752,13 +2777,15 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       if ('clear' in this.index && typeof this.index.clear === 'function') {
         await this.index.clear()
       } else {
-        // Recreate index if no clear method
+        // Recreate index using plugin factory when available
-        this.index = this.setupIndex()
+        this.index = this.createIndex()
       }
 
       // Recreate metadata index to clear cached data
-      // Bug: Metadata index cache was not being cleared, causing find() with type filters to return stale data
+      const clearMetadataFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('metadataIndex')
-      this.metadataIndex = new MetadataIndexManager(this.storage)
+      this.metadataIndex = clearMetadataFactory
+        ? clearMetadataFactory(this.storage)
+        : new MetadataIndexManager(this.storage)
       await this.metadataIndex.init()
 
       // Reset dimensions
@@ -2919,8 +2946,9 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       clone.storage.currentBranch = branchName
       // isInitialized inherited from prototype
 
-      // Shallow copy HNSW index (INSTANT - just copies Map references)
+      // Create HNSW index (uses plugin factory when available)
-      clone.index = this.setupIndex()
+      clone.index = this.createIndex()
+      clone.indexIsTypeAware = this.indexIsTypeAware
 
       // Enable COW (handle both HNSWIndex and TypeAwareHNSWIndex)
       if ('enableCOW' in clone.index && typeof clone.index.enableCOW === 'function') {
@@ -2928,7 +2956,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       }
 
       // Fast rebuild for small indexes from COW storage (Metadata/Graph are fast)
-      clone.metadataIndex = new MetadataIndexManager(clone.storage)
+      const cloneMetadataFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('metadataIndex')
+      clone.metadataIndex = cloneMetadataFactory
+        ? cloneMetadataFactory(clone.storage)
+        : new MetadataIndexManager(clone.storage)
       await clone.metadataIndex.init()
 
       // GraphAdjacencyIndex SINGLETON pattern for fork()
@@ -2937,7 +2968,13 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       // create a fresh instance for the clone's branch.
       ;(clone.storage as any).graphIndex = undefined
       ;(clone.storage as any).graphIndexPromise = undefined
+      const cloneGraphFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('graphIndex')
+      if (cloneGraphFactory) {
+        clone.graphIndex = cloneGraphFactory(clone.storage)
+        clone.storage.setGraphIndex(clone.graphIndex)
+      } else {
         clone.graphIndex = await (clone.storage as any).getGraphIndex()
+      }
       // getGraphIndex() will rebuild automatically if data exists (via _initializeGraphIndex)
 
       // Mark as initialized
@@ -3021,15 +3058,24 @@ export class Brainy<T = any> implements BrainyInterface<T> {
 
       // Fix: Reload indexes from new branch WITHOUT recreating storage
       // Previous implementation called init() which recreated storage, losing currentBranch
-      this.index = this.setupIndex()
+      this.index = this.createIndex()
-      this.metadataIndex = new (MetadataIndexManager as any)(this.storage)
+      const checkoutMetadataFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('metadataIndex')
+      this.metadataIndex = checkoutMetadataFactory
+        ? checkoutMetadataFactory(this.storage)
+        : new MetadataIndexManager(this.storage)
       await this.metadataIndex.init()
 
       // GraphAdjacencyIndex SINGLETON pattern for checkout()
       // Invalidate the old graphIndex (it has data from the old branch)
       // and get a fresh instance for the new branch
       ;(this.storage as any).invalidateGraphIndex()
+      const checkoutGraphFactory = this.pluginRegistry.getProvider<(storage: StorageAdapter) => any>('graphIndex')
+      if (checkoutGraphFactory) {
+        this.graphIndex = checkoutGraphFactory(this.storage)
+        this.storage.setGraphIndex(this.graphIndex)
+      } else {
         this.graphIndex = await (this.storage as any).getGraphIndex()
+      }
 
       // Reset lazy loading state when switching branches
       // Indexes contain data from previous branch, must rebuild for new branch
@@ -4264,6 +4310,94 @@ export class Brainy<T = any> implements BrainyInterface<T> {
     }
   }
 
+  /**
+   * Plugin and provider diagnostics — shows what's active and how subsystems are wired.
+   *
+   * @example
+   * ```typescript
+   * const diag = brain.diagnostics()
+   * console.log(diag.providers) // { hnsw: { source: 'plugin' }, ... }
+   * console.log(diag.indexes.graph.wiredToStorage) // true
+   * ```
+   */
+  diagnostics(): DiagnosticsResult {
+    const wellKnownKeys = [
+      'metadataIndex', 'graphIndex', 'entityIdMapper', 'cache',
+      'hnsw', 'roaring', 'embeddings', 'embedBatch', 'distance', 'msgpack'
+    ] as const
+
+    const providers: Record<string, { source: 'plugin' | 'default' }> = {}
+    for (const key of wellKnownKeys) {
+      providers[key] = {
+        source: this.pluginRegistry.hasProvider(key) ? 'plugin' : 'default'
+      }
+    }
+
+    const hnswSize = this.index.size()
+    const metadataInitialized = !!this.metadataIndex
+    const graphInitialized = !!this.graphIndex
+    const storageGraphIndex = (this.storage as any).graphIndex
+
+    return {
+      version: getBrainyVersion(),
+      plugins: {
+        active: this.pluginRegistry.getActivePlugins(),
+        count: this.pluginRegistry.getActivePlugins().length
+      },
+      providers,
+      indexes: {
+        hnsw: {
+          size: hnswSize,
+          type: this.index.constructor.name
+        },
+        metadata: {
+          type: this.metadataIndex?.constructor.name || 'none',
+          initialized: metadataInitialized
+        },
+        graph: {
+          type: this.graphIndex?.constructor.name || 'none',
+          initialized: graphInitialized,
+          wiredToStorage: graphInitialized && storageGraphIndex === this.graphIndex
+        }
+      }
+    }
+  }
+
+  /**
+   * Assert that specific providers are supplied by a plugin (not using JS fallback).
+   *
+   * Call after init() in production to fail fast if a paid plugin (e.g. cortex)
+   * isn't providing the expected acceleration. Throws if any listed key is using
+   * the default JavaScript implementation.
+   *
+   * @param keys - Provider keys that MUST come from a plugin
+   * @throws Error listing which providers are falling back to defaults
+   *
+   * @example
+   * ```typescript
+   * const brain = new Brainy()
+   * await brain.init()
+   *
+   * // Fail fast if cortex isn't providing these
+   * brain.requireProviders(['distance', 'embeddings', 'metadataIndex', 'graphIndex'])
+   * ```
+   */
+  requireProviders(keys: string[]): void {
+    const missing = keys.filter(k => !this.pluginRegistry.hasProvider(k))
+    if (missing.length > 0) {
+      const active = this.pluginRegistry.getActivePlugins()
+      const pluginInfo = active.length > 0
+        ? `Active plugins: ${active.join(', ')}`
+        : 'No plugins active'
+      throw new Error(
+        `[brainy] Required providers using JS fallback: ${missing.join(', ')}. ` +
+        `${pluginInfo}. ` +
+        `These providers must be supplied by a plugin for this deployment. ` +
+        `Check plugin installation, license, and native module availability.`
+      )
+    }
+  }
+
   /**
    * Efficient Pagination API - Production-scale pagination using index-first approach
    * Automatically optimizes based on query type and applies pagination at the index level
@@ -4655,7 +4789,16 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       return []
     }
 
-    // Use native WASM batch API: single forward pass instead of N individual calls
+    // Plugin provides native batch embedding — single forward pass for all texts
+    const batchProvider = this.pluginRegistry.getProvider<(texts: string[]) => Promise<number[][]>>('embedBatch')
+    if (batchProvider) {
+      return batchProvider(texts)
+    }
+    // Plugin provides single-text embedding engine — map through it
+    if (this.pluginRegistry.hasProvider('embeddings')) {
+      return Promise.all(texts.map(t => this.embedder(t)))
+    }
+    // Default: WASM batch API (single forward pass, more efficient than N calls)
     return await embeddingManager.embedBatch(texts, options)
   }
 
@@ -5481,9 +5624,9 @@ export class Brainy<T = any> implements BrainyInterface<T> {
     const vector = params.vector || (await this.embed(params.query!))
     const limit = params.limit || 10
 
-    // Phase 2: Pass type for TypeAwareHNSWIndex (10x faster for type-specific queries)
+    // Pass type for type-aware indexes (10x faster for type-specific queries)
-    const searchResults = this.index instanceof TypeAwareHNSWIndex
+    const searchResults: [string, number][] = this.indexIsTypeAware && params.type
-      ? await this.index.search(vector, limit * 2, params.type as any)
+      ? await (this.index as any).search(vector, limit * 2, params.type as any)
       : await this.index.search(vector, limit * 2)
 
     // Batch-load entities for 10-50x faster cloud storage performance
@@ -5512,9 +5655,9 @@ export class Brainy<T = any> implements BrainyInterface<T> {
     const nearEntity = await this.get(params.near.id)
     if (!nearEntity) return []
 
-    // Phase 2: Pass type for TypeAwareHNSWIndex
+    // Pass type for type-aware indexes
-    const nearResults = this.index instanceof TypeAwareHNSWIndex
+    const nearResults: [string, number][] = this.indexIsTypeAware && params.type
-      ? await this.index.search(nearEntity.vector, params.limit || 10, params.type as any)
+      ? await (this.index as any).search(nearEntity.vector, params.limit || 10, params.type as any)
       : await this.index.search(nearEntity.vector, params.limit || 10)
 
     // Filter by threshold first to minimize batch fetch
@@ -6068,11 +6211,16 @@ export class Brainy<T = any> implements BrainyInterface<T> {
       throw new Error('Brain must be initialized before warming up embeddings. Call init() first.')
     }
 
-    console.log('🚀 Warming up embedding engine...')
+    // Plugin-provided embeddings are already ready (native, no WASM warmup needed)
+    if (this.pluginRegistry.hasProvider('embeddings')) {
+      return
+    }
+
+    console.log('Warming up embedding engine...')
     const start = Date.now()
     await embeddingManager.init()
     const elapsed = Date.now() - start
-    console.log(`✅ Embedding engine ready in ${elapsed}ms`)
+    console.log(`Embedding engine ready in ${elapsed}ms`)
   }
 
   /**
@@ -6081,6 +6229,10 @@ export class Brainy<T = any> implements BrainyInterface<T> {
    * @returns true if embedding engine is ready for immediate use
    */
   isEmbeddingReady(): boolean {
+    // Plugin-provided embeddings are always ready (native, no WASM init required)
+    if (this.pluginRegistry.hasProvider('embeddings')) {
+      return true
+    }
     return embeddingManager.isInitialized()
   }
 
@@ -6177,6 +6329,23 @@ export class Brainy<T = any> implements BrainyInterface<T> {
     return new HNSWIndex(indexConfig as any, this.distance, { persistMode })
   }
 
+  /**
+   * Create an HNSW index, using plugin factory when available.
+   * Shared by init(), fork(), checkout(), and clear() to avoid duplication.
+   */
+  private createIndex(): HNSWIndex | TypeAwareHNSWIndex {
+    const hnswFactory = this.pluginRegistry.getProvider<(config: any, distance: DistanceFunction, options: any) => any>('hnsw')
+    if (hnswFactory) {
+      const persistMode = this.resolveHNSWPersistMode()
+      return hnswFactory(
+        { ...this.config.index, distanceFunction: this.distance },
+        this.distance,
+        { storage: this.storage, persistMode }
+      )
+    }
+    return this.setupIndex()
+  }
+
   /**
    * Resolve HNSW persistence mode.
    * Extracted so both setupIndex() and the HNSW plugin factory path can use it.

src/cli/commands/core.ts

@@ -918,5 +918,55 @@ export const coreCommands = {
       console.error(chalk.red(error.message))
       process.exit(1)
     }
+  },
+
+  /**
+   * Show plugin and provider diagnostics
+   */
+  async diagnostics(options: CoreOptions) {
+    try {
+      const brain = getBrainy()
+      await brain.init()
+
+      const diag = brain.diagnostics()
+
+      if (options.json) {
+        formatOutput(diag, options)
+        return
+      }
+
+      console.log(chalk.bold('\nBrainy Diagnostics'))
+      console.log(chalk.dim(`Version: ${diag.version}`))
+      console.log()
+
+      // Plugins
+      console.log(chalk.bold('Plugins:'))
+      if (diag.plugins.count === 0) {
+        console.log(chalk.dim('  (none active)'))
+      } else {
+        for (const name of diag.plugins.active) {
+          console.log(chalk.green(`  ✓ ${name}`))
+        }
+      }
+      console.log()
+
+      // Providers
+      console.log(chalk.bold('Providers:'))
+      for (const [key, info] of Object.entries(diag.providers)) {
+        const icon = info.source === 'plugin' ? chalk.green('✓ plugin') : chalk.dim('default')
+        console.log(`  ${key.padEnd(16)} ${icon}`)
+      }
+      console.log()
+
+      // Indexes
+      console.log(chalk.bold('Indexes:'))
+      console.log(`  HNSW:      ${diag.indexes.hnsw.type} (${diag.indexes.hnsw.size} vectors)`)
+      console.log(`  Metadata:  ${diag.indexes.metadata.type} (initialized: ${diag.indexes.metadata.initialized})`)
+      console.log(`  Graph:     ${diag.indexes.graph.type} (initialized: ${diag.indexes.graph.initialized}, wired: ${diag.indexes.graph.wiredToStorage})`)
+      console.log()
+    } catch (error: any) {
+      console.error(chalk.red('Diagnostics failed: ' + error.message))
+      process.exit(1)
+    }
   }
 }

src/cli/index.ts

@@ -174,6 +174,14 @@ program
   .option('-f, --format <format>', 'Output format (json|csv|jsonl)', 'json')
   .action(coreCommands.export)
 
+program
+  .command('diagnostics')
+  .alias('diag')
+  .description('Show plugin and provider diagnostics')
+  .option('--json', 'Output as JSON')
+  .option('--pretty', 'Pretty print JSON')
+  .action(coreCommands.diagnostics)
+
 // ===== Neural Commands =====
 
 program

src/index.ts

@@ -14,6 +14,9 @@ import { Brainy } from './brainy.js'
 
 export { Brainy }
 
+// Export diagnostics result type
+export type { DiagnosticsResult } from './brainy.js'
+
 // Export Brainy configuration and types
 export type {
   BrainyConfig,

src/neural/improvedNeuralAPI.ts

@@ -92,6 +92,7 @@ interface ItemWithMetadata {
 export class ImprovedNeuralAPI {
   private brain: any // Brainy instance
   private config: NeuralAPIConfig
+  private distanceFn: (a: Vector, b: Vector) => number
 
   // Caching for performance
   private similarityCache = new Map<string, number | SimilarityResult>()
@@ -104,6 +105,7 @@ export class ImprovedNeuralAPI {
 
   constructor(brain: any, config: NeuralAPIConfig = {}) {
     this.brain = brain
+    this.distanceFn = brain.distance || cosineDistance
     this.config = {
       cacheSize: 1000,
       defaultAlgorithm: 'auto',
@@ -1458,7 +1460,7 @@ export class ImprovedNeuralAPI {
 
     switch (metric) {
       case 'cosine':
-        score = 1 - cosineDistance(v1, v2)
+        score = 1 - this.distanceFn(v1, v2)
         break
       case 'euclidean':
         score = 1 / (1 + euclideanDistance(v1, v2))
@@ -1467,7 +1469,7 @@ export class ImprovedNeuralAPI {
         score = 1 / (1 + this._manhattanDistance(v1, v2))
         break
       default:
-        score = 1 - cosineDistance(v1, v2)
+        score = 1 - this.distanceFn(v1, v2)
     }
 
     if (options.detailed) {
@@ -3047,7 +3049,7 @@ export class ImprovedNeuralAPI {
                   continue
                 }
 
-                const similarity = 1 - cosineDistance(
+                const similarity = 1 - this.distanceFn(
                   Array.from(c1.centroid) as number[],
                   Array.from(c2.centroid) as number[]
                 )

src/neural/neuralAPI.ts

@@ -132,12 +132,14 @@ export interface LODConfig {
  */
 export class NeuralAPI {
   private brain: any // Brainy instance
+  private distanceFn: (a: Vector, b: Vector) => number
   private similarityCache: Map<string, number> = new Map()
   private clusterCache: Map<string, any> = new Map() // Enhanced for enterprise
   private hierarchyCache: Map<string, SemanticHierarchy> = new Map()
 
   constructor(brain: any) {
     this.brain = brain
+    this.distanceFn = brain.distance || cosineDistance
   }
   
   // ===== SMART USER-FRIENDLY API =====
@@ -471,7 +473,7 @@ export class NeuralAPI {
     }
     
     // Calculate similarity
-    const score = cosineDistance(itemA.vector, itemB.vector)
+    const score = this.distanceFn(itemA.vector, itemB.vector)
     
     this.similarityCache.set(cacheKey, score)
     
@@ -498,7 +500,7 @@ export class NeuralAPI {
   }
   
   private async similarityByVector(vectorA: Vector, vectorB: Vector, options?: SimilarityOptions): Promise<number | SimilarityResult> {
-    const score = cosineDistance(vectorA, vectorB)
+    const score = this.distanceFn(vectorA, vectorB)
     
     if (options?.explain) {
       return {
@@ -610,7 +612,7 @@ export class NeuralAPI {
         // Find minimum distance to existing sample
         let minDistance = Infinity
         for (const selected of sample) {
-          const distance = cosineDistance(candidate.vector, selected.vector)
+          const distance = this.distanceFn(candidate.vector, selected.vector)
           minDistance = Math.min(minDistance, distance)
         }
         
@@ -652,7 +654,7 @@ export class NeuralAPI {
         let bestDistance = Infinity
         
         for (let c = 0; c < k; c++) {
-          const distance = cosineDistance(item.vector, centroids[c])
+          const distance = this.distanceFn(item.vector, centroids[c])
           if (distance < bestDistance) {
             bestDistance = distance
             bestCluster = c
@@ -679,7 +681,7 @@ export class NeuralAPI {
         let bestDistance = Infinity
         
         for (let cc = 0; cc < k; cc++) {
-          const distance = cosineDistance(item.vector, centroids[cc])
+          const distance = this.distanceFn(item.vector, centroids[cc])
           if (distance < bestDistance) {
             bestDistance = distance
             bestCluster = cc
@@ -750,7 +752,7 @@ export class NeuralAPI {
       let merged = false
       
       for (let i = 0; i < result.length; i++) {
-        const similarity = cosineDistance(result[i].centroid, batchCluster.centroid)
+        const similarity = this.distanceFn(result[i].centroid, batchCluster.centroid)
         
         if (similarity > 0.8) {
           // Merge clusters

src/neural/signals/VerbEmbeddingSignal.ts

@@ -64,6 +64,7 @@ interface HistoricalEntry {
  */
 export class VerbEmbeddingSignal {
   private brain: Brainy
+  private distanceFn: (a: Vector, b: Vector) => number
   private options: Required<VerbEmbeddingSignalOptions>
 
   // Pre-computed verb type embeddings (loaded once at startup)
@@ -88,6 +89,7 @@ export class VerbEmbeddingSignal {
 
   constructor(brain: Brainy, options?: VerbEmbeddingSignalOptions) {
     this.brain = brain
+    this.distanceFn = (brain as any).distance || cosineDistance
     this.options = {
       minConfidence: options?.minConfidence ?? 0.60,
       minSimilarity: options?.minSimilarity ?? 0.55,
@@ -142,7 +144,7 @@ export class VerbEmbeddingSignal {
       const similarities: Array<{ type: VerbType; similarity: number }> = []
 
       for (const [verbType, typeEmbedding] of this.verbTypeEmbeddings) {
-        const distance = cosineDistance(embedding, typeEmbedding)
+        const distance = this.distanceFn(embedding, typeEmbedding)
         const similarity = 1 - distance  // Convert distance to similarity
         similarities.push({ type: verbType, similarity })
       }

src/plugin.ts

@@ -43,7 +43,8 @@ export interface BrainyPluginContext {
    * - 'cache'              — UnifiedCache replacement
    * - 'hnsw'               — HNSWIndex replacement
    * - 'roaring'            — RoaringBitmap32 replacement
-   * - 'embeddings'         — Embedding engine replacement
+   * - 'embeddings'         — Embedding engine replacement (single text)
+   * - 'embedBatch'         — Batch embedding engine (texts[] → vectors[])
    * - 'distance'           — Distance function overrides
    * - 'msgpack'            — Msgpack encode/decode
    *

src/storage/baseStorage.ts

@@ -379,6 +379,16 @@ export abstract class BaseStorage extends BaseStorageAdapter {
     }
   }
 
+  /**
+   * Set the graph index instance (used by Brainy to wire plugin-provided graph indexes).
+   * This ensures getVerbsBySource() uses the fast GraphAdjacencyIndex path
+   * instead of falling back to O(n) shard iteration.
+   */
+  public setGraphIndex(index: GraphAdjacencyIndex): void {
+    this.graphIndex = index
+    this.graphIndexPromise = Promise.resolve(index)
+  }
+
   /**
    * Ensure the storage adapter is initialized
    */

src/vfs/PathResolver.ts

@@ -201,9 +201,8 @@ export class PathResolver {
    * Falls back to graph traversal if MetadataIndex unavailable
    */
   private async resolveWithMetadataIndex(path: string): Promise<string> {
-    // Access MetadataIndexManager from brain's storage
+    // Access MetadataIndexManager from brain instance (not storage — metadataIndex lives on Brainy)
-    const storage = (this.brain as any).storage
+    const metadataIndex = (this.brain as any).metadataIndex
-    const metadataIndex = storage?.metadataIndex
 
     if (!metadataIndex) {
       // MetadataIndex not available, use graph traversal

src/vfs/semantic/SemanticPathResolver.ts

@@ -17,7 +17,7 @@ import { PathResolver } from '../PathResolver.js'
 import { VFSEntity, VFSError, VFSErrorCode } from '../types.js'
 import { SemanticPathParser, ParsedSemanticPath } from './SemanticPathParser.js'
 import { ProjectionRegistry } from './ProjectionRegistry.js'
-import { UnifiedCache } from '../../utils/unifiedCache.js'
+import { getGlobalCache, UnifiedCache } from '../../utils/unifiedCache.js'
 
 /**
  * Semantic Path Resolver
@@ -44,12 +44,8 @@ export class SemanticPathResolver {
     this.registry = registry
     this.parser = new SemanticPathParser()
 
-    // Use Brainy's UnifiedCache for semantic path caching
+    // Use global UnifiedCache (picks up plugin-provided cache when available)
-    // Zero-config: Uses 2GB default from UnifiedCache
+    this.cache = getGlobalCache()
-    this.cache = new UnifiedCache({
-      enableRequestCoalescing: true,
-      enableFairnessCheck: true
-    })
 
     // Create traditional path resolver (uses its own optimized cache with defaults)
     this.pathResolver = new PathResolver(brain, rootEntityId)