CRITICAL FIX: Single blob read error no longer corrupts entire VFS tree Root Causes Fixed: 1. Uncaught blob errors in readFile() triggered VFS re-initialization 2. Race conditions in initializeRoot() created duplicate roots 3. Wrong root selection algorithm (oldest vs most children) Architectural Solution: - **Error Isolation**: Blob errors caught and re-thrown as VFSError VFS tree structure completely isolated from file content errors (src/vfs/VirtualFileSystem.ts:288-321) - **Singleton Promise Pattern**: Prevents duplicate root creation Concurrent init() calls wait for same initialization promise Acts as automatic mutex without custom lock class (src/vfs/VirtualFileSystem.ts:180-199) - **Smart Root Selection**: Selects root with MOST children (not oldest) Auto-heals existing duplicates on init() Logs cleanup suggestions for empty roots (src/vfs/VirtualFileSystem.ts:283-334) Production Impact: - Workshop production: 5 duplicate roots, 1,836 files orphaned - After fix: Zero duplicate roots possible, auto-healing - All 100 VFS tests pass ✅ Additional Fix: Remove Sharp native dependency (v5.8.0) ImageHandler rewritten using pure JavaScript: - exifr (already installed) for EXIF extraction - probe-image-size for image dimensions/format - Zero native dependencies (removed 10MB of native binaries) - All 25 image handler tests pass ✅ - No more test crashes from Sharp/libvips worker thread issues Test Results: - VFS tests: 100/100 pass ✅ - Image handler tests: 25/25 pass ✅ - Overall: 1157/1200 tests pass (18 pre-existing timeout issues) - Build: Successful, zero TypeScript errors ✅ Features Complete (TIER 1 - v5.8.0): - Transaction system (36 unit + 35 integration tests) - Duplicate check optimization (O(n) → O(log n)) - GraphIndex pagination - Comprehensive filter documentation |
||
|---|---|---|
| .. | ||
| discovery | ||
| display | ||
| intelligentImport | ||
| typeMatching | ||
| apiServerAugmentation.ts | ||
| auditLogAugmentation.ts | ||
| AugmentationMetadataContract.ts | ||
| batchProcessingAugmentation.ts | ||
| brainyAugmentation.ts | ||
| cacheAugmentation.ts | ||
| conduitAugmentations.ts | ||
| configResolver.ts | ||
| connectionPoolAugmentation.ts | ||
| defaultAugmentations.ts | ||
| discovery.ts | ||
| entityRegistryAugmentation.ts | ||
| indexAugmentation.ts | ||
| intelligentVerbScoringAugmentation.ts | ||
| manifest.ts | ||
| metadataEnforcer.ts | ||
| metricsAugmentation.ts | ||
| monitoringAugmentation.ts | ||
| neuralImport.ts | ||
| rateLimitAugmentation.ts | ||
| README.md | ||
| requestDeduplicatorAugmentation.ts | ||
| storageAugmentation.ts | ||
| storageAugmentations.ts | ||
| synapseAugmentation.ts | ||
| universalDisplayAugmentation.ts | ||
| versioningAugmentation.ts | ||
Brainy Augmentations
This directory contains the augmentation implementations for Brainy. Augmentations are pluggable components that extend Brainy's functionality in various ways.
Available Augmentations
Core Augmentations
IntelligentImportAugmentation
Automatically detects and processes CSV, Excel, and PDF files with intelligent extraction. This augmentation is enabled by default and provides:
- CSV Support: Auto-detection of encoding, delimiters, and field types
- Excel Support: Multi-sheet extraction with metadata preservation
- PDF Support: Text extraction, table detection, and metadata extraction
- Type Inference: Automatically infers data types (string, number, boolean, date)
- Neural Integration: Seamlessly integrates with entity extraction and relationship detection
import { Brainy } from '@soulcraft/brainy'
const brain = new Brainy({
intelligentImport: {
enableCSV: true,
enableExcel: true,
enablePDF: true,
maxFileSize: 100 * 1024 * 1024 // 100MB
}
})
await brain.init()
// Import CSV with auto-detection
await brain.import('customers.csv')
// Import Excel with specific sheets
await brain.import('sales-data.xlsx', {
excelSheets: ['Q1', 'Q2']
})
// Import PDF with table extraction
await brain.import('report.pdf', {
pdfExtractTables: true
})
See: Import Anything Guide | Example
Conduit Augmentations
Conduit augmentations provide data synchronization between Brainy instances.
WebSocketConduitAugmentation
A conduit augmentation that syncs Brainy instances using WebSockets. This is used for syncing between browsers and servers, or between servers.
import { createConduitAugmentation, augmentationPipeline } from '@soulcraft/brainy'
// Create a WebSocket conduit augmentation
const wsConduit = await createConduitAugmentation('websocket', 'my-websocket-sync')
// Register the augmentation with the pipeline
augmentationPipeline.register(wsConduit)
// Connect to another Brainy instance
const connectionResult = await wsConduit.establishConnection(
'wss://your-websocket-server.com/brainy-sync',
{ protocols: 'brainy-sync' }
)
WebRTCConduitAugmentation
A conduit augmentation that syncs Brainy instances using WebRTC. This is used for direct peer-to-peer syncing between browsers.
import { createConduitAugmentation, augmentationPipeline } from '@soulcraft/brainy'
// Create a WebRTC conduit augmentation
const webrtcConduit = await createConduitAugmentation('webrtc', 'my-webrtc-sync')
// Register the augmentation with the pipeline
augmentationPipeline.register(webrtcConduit)
// Connect to a peer
const connectionResult = await webrtcConduit.establishConnection(
'peer-id-to-connect-to',
{
signalServerUrl: 'wss://your-signal-server.com',
localPeerId: 'my-peer-id',
iceServers: [{ urls: 'stun:stun.l.google.com:19302' }]
}
)
ServerSearchConduitAugmentation
A specialized conduit augmentation that provides functionality for searching a server-hosted Brainy instance and storing results locally. This allows you to:
- Search a server-hosted Brainy instance from a browser
- Store the search results in a local Brainy instance
- Perform further searches against the local instance without needing to query the server again
- Add data to both local and server instances
import {
ServerSearchConduitAugmentation,
createServerSearchAugmentations,
augmentationPipeline
} from '@soulcraft/brainy'
// Using the factory function (recommended)
const { conduit, activation, connection } = await createServerSearchAugmentations(
'wss://your-brainy-server.com/ws',
{ protocols: 'brainy-sync' }
)
// Register the augmentations with the pipeline
augmentationPipeline.register(conduit)
augmentationPipeline.register(activation)
// Search the server and store results locally
const serverSearchResult = await conduit.searchServer(
connection.connectionId,
'your search query',
5 // limit
)
// Search the local instance
const localSearchResult = await conduit.searchLocal('your search query', 5)
// Perform a combined search (local first, then server if needed)
const combinedSearchResult = await conduit.searchCombined(
connection.connectionId,
'your search query',
5
)
// Add data to both local and server
const addResult = await conduit.addToBoth(
connection.connectionId,
'Text to add',
{ /* metadata */ }
)
Activation Augmentations
Activation augmentations dictate how Brainy initiates actions, responses, or data manipulations.
ServerSearchActivationAugmentation
An activation augmentation that provides actions for server search functionality. This works in conjunction with the ServerSearchConduitAugmentation to provide a complete solution for browser-server search.
import {
ServerSearchActivationAugmentation,
createServerSearchAugmentations,
augmentationPipeline
} from '@soulcraft/brainy'
// Using the factory function (recommended)
const { conduit, activation, connection } = await createServerSearchAugmentations(
'wss://your-brainy-server.com/ws',
{ protocols: 'brainy-sync' }
)
// Register the augmentations with the pipeline
augmentationPipeline.register(conduit)
augmentationPipeline.register(activation)
// Use the activation augmentation to search the server
const serverSearchAction = activation.triggerAction('searchServer', {
connectionId: connection.connectionId,
query: 'your search query',
limit: 5
})
if (serverSearchAction.success) {
// The data property contains a promise that will resolve to the search results
const serverSearchResult = await serverSearchAction.data
console.log('Server search results:', serverSearchResult)
}
// Other available actions:
// - 'connectToServer': Connect to a server
// - 'searchLocal': Search the local instance
// - 'searchCombined': Search both local and server
// - 'addToBoth': Add data to both local and server
Using the Augmentation Pipeline
The augmentation pipeline provides a way to execute augmentations based on their type.
import { augmentationPipeline } from '@soulcraft/brainy'
// Execute a conduit augmentation
const conduitResults = await augmentationPipeline.executeConduitPipeline(
'methodName',
[arg1, arg2, ...],
{ /* options */ }
)
// Execute an activation augmentation
const activationResults = await augmentationPipeline.executeActivationPipeline(
'methodName',
[arg1, arg2, ...],
{ /* options */ }
)
Creating Custom Augmentations
To create a custom augmentation, implement one of the augmentation interfaces:
ISenseAugmentation: For processing raw dataIConduitAugmentation: For data synchronizationICognitionAugmentation: For reasoning and inferenceIMemoryAugmentation: For data storageIPerceptionAugmentation: For data interpretation and visualizationIDialogAugmentation: For natural language processingIActivationAugmentation: For triggering actions
Example:
import { AugmentationType, IActivationAugmentation } from '@soulcraft/brainy'
class MyCustomActivation implements IActivationAugmentation {
readonly
name = 'my-custom-activation'
readonly
description = 'My custom activation augmentation'
enabled = true
getType(): AugmentationType {
return AugmentationType.ACTIVATION
}
async initialize(): Promise<void> {
// Initialization code
}
async shutDown(): Promise<void> {
// Cleanup code
}
async getStatus(): Promise<'active' | 'inactive' | 'error'> {
return 'active'
}
triggerAction(actionName: string, parameters
?:
Record<string, unknown>
):
AugmentationResponse<unknown> {
// Implementation
}
generateOutput(knowledgeId: string, format: string): AugmentationResponse<string | Record<string, unknown
>> {
// Implementation
}
interactExternal(systemId
:
string, payload
:
Record < string, unknown >
):
AugmentationResponse < unknown > {
// Implementation
}
}