@oculum/scanner 1.0.11 → 1.0.13

package/src/{layer2/ai-rag-safety.ts → detect/ai-code/rag-safety.ts}

@@ -9,7 +9,8 @@
  * - Context logging risks
  */
 
-import type { Vulnerability, VulnerabilitySeverity, VulnerabilityCategory } from '../types'
+import type { Vulnerability, VulnerabilitySeverity, VulnerabilityCategory } from '../../shared/types'
+import type { ParsedFile } from '../../shared/parsed-file'
 import {
   isComment,
   isTestOrMockFile,
@@ -17,7 +18,9 @@ import {
   isScannerOrFixtureFile,
   isExampleDirectory,
   isLibraryCode,
-} from '../utils/context-helpers'
+} from '../../parse/file-classifier'
+
+const BASE_CONFIDENCE = 0.45
 
 // ============================================================================
 // Context Detection
@@ -598,6 +601,85 @@ const EMBEDDING_POISONING_PATTERNS: RAGSafetyPattern[] = [
   },
 ]
 
+/**
+ * Phase 6 Task 4: Cross-Tenant RAG Detection Patterns
+ * Detect shared vector stores without tenant filtering that could leak data
+ */
+const CROSS_TENANT_PATTERNS: RAGSafetyPattern[] = [
+  // Shared vector store without tenant filter
+  {
+    name: 'Shared vector store without tenant filter',
+    pattern: /\b(?:vectorStore|index|collection)\s*=\s*(?:new\s+)?(?:PineconeStore|ChromaDB|Weaviate|Qdrant|Milvus|PGVector|VectorStore)\s*\([^)]*\)(?![\s\S]{0,100}(?:filter|where|tenant|user|org|namespace))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'high',
+    description: 'Vector store initialized without tenant filtering. In multi-tenant applications, this could leak data across tenants.',
+    suggestedFix: 'Always include tenant/user ID in vector store configuration or filters: new VectorStore({ namespace: tenantId })',
+  },
+  // Query without tenant in metadata filter
+  {
+    name: 'Vector query missing tenant filter',
+    pattern: /\.(?:query|search|similaritySearch)\s*\(\s*[^,)]+(?:,\s*\{[^}]*\})?\s*\)(?![\s\S]{0,80}(?:tenantId|tenant_id|orgId|org_id|userId|user_id|namespace))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'high',
+    description: 'Vector store query without tenant filtering. Results may include documents from other tenants.',
+    suggestedFix: 'Add tenant filter to query: .query(q, { filter: { tenantId: ctx.tenant.id } })',
+  },
+  // Global index access pattern
+  {
+    name: 'Global index without scoping',
+    pattern: /(?:const|let|var)\s+(?:index|vectorIndex|searchIndex)\s*=\s*(?:await\s+)?(?:getIndex|loadIndex|connectIndex|initializeIndex)\s*\(\s*(?:['"`][^'"`]+['"`])?\s*\)(?![\s\S]{0,50}(?:tenant|user|org|scope))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'medium',
+    description: 'Global index loaded without tenant scoping. Consider using tenant-specific indexes or namespaces.',
+    suggestedFix: 'Use tenant-scoped index: const index = await getIndex(tenantId) or use namespace parameter',
+  },
+  // Multi-tenant store without isolation
+  {
+    name: 'Multi-tenant store missing isolation',
+    pattern: /(?:multiTenant|shared|global)(?:Store|Index|Collection)\s*\.(?:query|search|add|upsert)(?![\s\S]{0,80}(?:tenantId|tenant_id|isolate|partition|namespace))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'critical',
+    description: 'Multi-tenant store accessed without tenant isolation. Data from all tenants is accessible.',
+    suggestedFix: 'Always pass tenant identifier: multiTenantStore.query(q, { tenantId })',
+  },
+  // Embedding documents without tenant metadata
+  {
+    name: 'Document embedded without tenant metadata',
+    pattern: /\.(?:addDocuments|upsert|add)\s*\(\s*\[?\s*\{[^}]*(?:content|text|pageContent)[^}]*\}(?![^}]*(?:tenantId|tenant_id|orgId|organizationId|userId|user_id))/gi,
+    riskType: 'corpus_poisoning',
+    baseSeverity: 'high',
+    description: 'Documents embedded without tenant metadata. Without tenant ID, documents cannot be filtered per-tenant.',
+    suggestedFix: 'Include tenant ID in document metadata: { content, metadata: { tenantId: ctx.tenant.id } }',
+  },
+  // Retriever without tenant context
+  {
+    name: 'Retriever created without tenant context',
+    pattern: /(?:asRetriever|createRetriever|getRetriever)\s*\(\s*(?:\{[^}]*\})?\s*\)(?![\s\S]{0,80}(?:filter|tenant|user|org|metadata))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'medium',
+    description: 'Retriever created without tenant filtering configuration. Retrieved documents may cross tenant boundaries.',
+    suggestedFix: 'Configure retriever with tenant filter: vectorStore.asRetriever({ filter: { tenantId } })',
+  },
+  // Semantic search across all tenants
+  {
+    name: 'Semantic search without tenant restriction',
+    pattern: /semanticSearch\s*\(\s*[^,)]+\s*\)(?![\s\S]{0,50}(?:tenant|user|org|filter|where|scope))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'high',
+    description: 'Semantic search without tenant restriction. Search results span all tenants.',
+    suggestedFix: 'Add tenant restriction: semanticSearch(query, { tenantId: ctx.tenant.id })',
+  },
+  // RAG chain without tenant context
+  {
+    name: 'RAG chain missing tenant context',
+    pattern: /(?:createRetrievalChain|RetrievalQAChain|ConversationalRetrievalChain)\.(?:fromLLM|from)?\s*\([^)]*\)(?![\s\S]{0,100}(?:filter|tenant|user|metadata))/gi,
+    riskType: 'unscoped_retrieval',
+    baseSeverity: 'medium',
+    description: 'RAG chain created without tenant context. Chain may retrieve documents from all tenants.',
+    suggestedFix: 'Pass tenant-filtered retriever to chain or add metadata filtering',
+  },
+]
+
 /**
  * Chunk Boundary Exploitation Patterns
  * Detects cross-chunk injection vulnerabilities
@@ -679,7 +761,8 @@ function mapRiskTypeToCategory(riskType: RAGSafetyPattern['riskType']): Vulnerab
  */
 export function detectRAGSafetyIssues(
   content: string,
-  filePath: string
+  filePath: string,
+  options?: { parsed?: ParsedFile }
 ): Vulnerability[] {
   const vulnerabilities: Vulnerability[] = []
 
@@ -692,7 +775,7 @@ export function detectRAGSafetyIssues(
     return vulnerabilities
   }
 
-  const lines = content.split('\n')
+  const lines = options?.parsed?.lines ?? content.split('\n')
   const isTestFile = isTestOrMockFile(filePath)
   const isExample = isExampleDirectory(filePath)
   const isLibrary = isLibraryCode(filePath)
@@ -710,6 +793,8 @@ export function detectRAGSafetyIssues(
     ...QUERY_INJECTION_PATTERNS,
     ...EMBEDDING_POISONING_PATTERNS,
     ...CHUNK_INJECTION_PATTERNS,
+    // Phase 6: Cross-tenant detection
+    ...CROSS_TENANT_PATTERNS,
   ]
 
   for (const pattern of allPatterns) {
@@ -878,7 +963,9 @@ export function detectRAGSafetyIssues(
         suggestedFix: pattern.suggestedFix,
         confidence: severity === 'info' ? 'low' : 'medium',
         layer: 2,
+        source: 'ai_code' as const,
         requiresAIValidation: severity !== 'info' && pattern.riskType !== 'context_logging',
+        baseConfidence: BASE_CONFIDENCE,
       })
     }
   }

package/src/{layer2/ai-schema-validation.ts → detect/ai-code/schema-validation.ts}

@@ -9,13 +9,16 @@
  * - Tool invocation parameters not validated
  */
 
-import type { Vulnerability, VulnerabilitySeverity } from '../types'
+import type { Vulnerability, VulnerabilitySeverity } from '../../shared/types'
+import type { ParsedFile } from '../../shared/parsed-file'
 import {
   isComment,
   isTestOrMockFile,
   isDocumentationFile,
   isScannerOrFixtureFile,
-} from '../utils/context-helpers'
+} from '../../parse/file-classifier'
+
+const BASE_CONFIDENCE = 0.40
 
 // ============================================================================
 // Context Detection
@@ -304,7 +307,8 @@ const TOOL_PARAM_PATTERNS: SchemaValidationPattern[] = [
  */
 export function detectAISchemaValidation(
   content: string,
-  filePath: string
+  filePath: string,
+  options?: { parsed?: ParsedFile }
 ): Vulnerability[] {
   const vulnerabilities: Vulnerability[] = []
 
@@ -317,7 +321,7 @@ export function detectAISchemaValidation(
     return vulnerabilities
   }
 
-  const lines = content.split('\n')
+  const lines = options?.parsed?.lines ?? content.split('\n')
   const isTestFile = isTestOrMockFile(filePath)
 
   // Check file-level validation presence
@@ -409,7 +413,9 @@ export function detectAISchemaValidation(
         suggestedFix: pattern.suggestedFix,
         confidence: severity === 'info' ? 'low' : 'medium',
         layer: 2,
+        source: 'ai_code' as const,
         requiresAIValidation: true, // Tier B - always validate with AI
+        baseConfidence: BASE_CONFIDENCE,
       })
     }
   }