@harness-engineering/graph 0.6.0 → 0.7.0

package/dist/index.d.mts

@@ -512,6 +512,33 @@ declare class RequirementIngestor {
     private linkByKeywordOverlap;
 }
 
+/**
+ * Domain inference for graph nodes.
+ *
+ * Precedence (highest to lowest):
+ *   1. node.metadata.domain (explicit)
+ *   2. extraPatterns (config-provided)
+ *   3. DEFAULT_PATTERNS (built-in)
+ *   4. Generic first non-blocklisted path segment
+ *   5. KnowledgeLinker connector source (path-less facts)
+ *   6. 'unknown'
+ *
+ * Pattern format: 'prefix/<dir>'. Single-segment prefix only. <dir> captures
+ * the segment immediately after the prefix.
+ */
+interface DomainInferenceOptions {
+    /** Additional patterns beyond the built-in defaults. Format: 'prefix/<dir>'. */
+    extraPatterns?: readonly string[];
+    /** Additional blocklisted segments beyond the built-in defaults. */
+    extraBlocklist?: readonly string[];
+}
+declare const DEFAULT_PATTERNS: readonly string[];
+declare const DEFAULT_BLOCKLIST: ReadonlySet<string>;
+declare function inferDomain(node: {
+    path?: string;
+    metadata?: Record<string, unknown>;
+}, options?: DomainInferenceOptions): string;
+
 type DriftClassification = 'new' | 'drifted' | 'stale' | 'contradicting';
 interface KnowledgeSnapshotEntry {
     readonly id: string;
@@ -591,7 +618,8 @@ interface AggregateResult {
 }
 declare class KnowledgeStagingAggregator {
     private readonly projectDir;
-    constructor(projectDir: string);
+    private readonly inferenceOptions;
+    constructor(projectDir: string, inferenceOptions?: DomainInferenceOptions);
     aggregate(extractorResults: readonly StagedEntry[], linkerResults: readonly StagedEntry[], diagramResults: readonly StagedEntry[]): Promise<AggregateResult>;
     private extractDocName;
     generateGapReport(knowledgeDir: string, store?: GraphStore): Promise<GapReport>;
@@ -691,8 +719,9 @@ interface CoverageReport {
     readonly generatedAt: string;
 }
 declare class CoverageScorer {
+    private readonly inferenceOptions;
+    constructor(inferenceOptions?: DomainInferenceOptions);
     score(store: GraphStore): CoverageReport;
-    private domainFromPath;
 }
 
 /**
@@ -725,7 +754,8 @@ interface SkippedEntry {
 }
 declare class KnowledgeDocMaterializer {
     private readonly store;
-    constructor(store: GraphStore);
+    private readonly inferenceOptions;
+    constructor(store: GraphStore, inferenceOptions?: DomainInferenceOptions);
     materialize(gapEntries: readonly GapEntry[], options: MaterializeOptions): Promise<MaterializeResult>;
     private resolveEntry;
     inferDomain(node: GraphNode): string | null;
@@ -758,6 +788,13 @@ interface KnowledgePipelineOptions {
     readonly analyzeImages?: boolean;
     readonly analysisProvider?: AnalysisProvider;
     readonly imagePaths?: readonly string[];
+    /**
+     * Domain-inference overrides threaded into KnowledgeStagingAggregator,
+     * CoverageScorer, and KnowledgeDocMaterializer. Sourced by the CLI from
+     * `harness.config.json#knowledge.domainPatterns` (→ extraPatterns) and
+     * `knowledge.domainBlocklist` (→ extraBlocklist). Defaults to {} when absent.
+     */
+    readonly inferenceOptions?: DomainInferenceOptions;
 }
 interface ExtractionCounts {
     readonly codeSignals: number;
@@ -782,6 +819,8 @@ interface KnowledgePipelineResult {
 declare class KnowledgePipelineRunner {
     private readonly store;
     constructor(store: GraphStore);
+    /** Resolved per-`run()` inference options. Set on entry to `run()`. */
+    private inferenceOptions;
     run(options: KnowledgePipelineOptions): Promise<KnowledgePipelineResult>;
     /** Run the remediation convergence loop; returns iteration count and accumulated materialization. */
     private runRemediationLoop;
@@ -1827,4 +1866,4 @@ declare class CascadeSimulator {
 
 declare const VERSION = "0.6.0";
 
-export { ALL_EXTRACTORS, type AggregateResult, type AnomalyDetectionOptions, type AnomalyReport, ApiPathExtractor, type ArticulationPoint, type AskGraphResult, type AssembledContext, Assembler, BusinessKnowledgeIngestor, CIConnector, CURRENT_SCHEMA_VERSION, type PackedEnvelope as CacheableEnvelope, type CascadeLayer, type CascadeNode, type CascadeResult, type CascadeSimulationOptions, CascadeSimulator, type ClassificationResult, CodeIngestor, CompositeProbabilityStrategy, type ConflictDetail, type ConflictPrediction, ConflictPredictor, type ConflictSeverity, type ConflictType, ConfluenceConnector, type ConnectorConfig, ContextQL, type ContextQLParams, type ContextQLResult, type Contradiction, ContradictionDetector, type ContradictionEntry, type ContradictionResult, type CoverageReport, CoverageScorer, D2Parser, DecisionIngestor, DesignConstraintAdapter, DesignIngestor, type DesignStrictness, type DesignViolation, type DetectedElement, type DiagramEntity, type DiagramFormatParser, type DiagramParseResult, DiagramParser, type DiagramRelationship, type DomainCoverage, type DomainCoverageScore, type DriftClassification, type DriftDetector, type DriftFinding, type DriftResult, EDGE_TYPES, type EdgeQuery, type EdgeType, EntityExtractor, EntityResolver, EnumConstantExtractor, type ExtractionCounts, type ExtractionRecord, ExtractionRunner, FigmaConnector, FusionLayer, type FusionResult, type GapEntry, type GapReport, GitIngestor, type GitRunner, GraphAnomalyAdapter, type GraphBudget, GraphComplexityAdapter, type GraphComplexityHotspot, type GraphComplexityResult, type GraphConnector, GraphConstraintAdapter, GraphCouplingAdapter, type GraphCouplingFileData, type GraphCouplingResult, type GraphCoverageReport, type GraphDeadCodeData, type GraphDependencyData, type GraphDriftData, type GraphEdge, GraphEdgeSchema, GraphEntropyAdapter, GraphFeedbackAdapter, type GraphFilterResult, type GraphHarnessCheckData, type GraphImpactData, type GraphLayerViolation, type GraphMetadata, type GraphNode, GraphNodeSchema, type GraphSnapshotSummary, type GraphStabilityTier, GraphStore, type HttpClient, INTENTS, ImageAnalysisExtractor, type ImageAnalysisExtractorOptions, type AnalysisProvider as ImageAnalysisProvider, type ImageAnalysisResult, type ImpactGroups, type IndependenceCheckParams, type IndependenceResult, type IngestResult, type Intent, IntentClassifier, JiraConnector, KnowledgeDocMaterializer, KnowledgeIngestor, type KnowledgePipelineOptions, type KnowledgePipelineResult, KnowledgePipelineRunner, type KnowledgeSnapshot, type KnowledgeSnapshotEntry, KnowledgeStagingAggregator, type Language, type LinkResult, type LoadGraphResult, type MaterializeOptions, type MaterializeResult, type MaterializedDoc, MermaidParser, MiroConnector, NODE_STABILITY, NODE_TYPES, type NodeCategory, type NodeQuery, type NodeType, OBSERVABILITY_TYPES, type OverlapDetail, PackedSummaryCache, type PairResult, PlantUmlParser, type ProbabilityStrategy, type ProjectionSpec, type RequirementCoverage, RequirementIngestor, type ResolvedEntity, ResponseFormatter, type SignalExtractor, type SkippedEntry, SlackConnector, type SourceLocation, type StagedEntry, type StatisticalOutlier, StructuralDriftDetector, SyncManager, type SyncMetadata, type TaskDefinition, TaskIndependenceAnalyzer, TestDescriptionExtractor, TopologicalLinker, type TraceabilityOptions, type TraceabilityResult, type TracedFile, VERSION, ValidationRuleExtractor, type VectorSearchResult, VectorStore, askGraph, classifyNodeCategory, createExtractionRunner, detectLanguage, groupNodesByImpact, linkToCode, loadGraph, normalizeIntent, project, queryTraceability, saveGraph };
+export { ALL_EXTRACTORS, type AggregateResult, type AnomalyDetectionOptions, type AnomalyReport, ApiPathExtractor, type ArticulationPoint, type AskGraphResult, type AssembledContext, Assembler, BusinessKnowledgeIngestor, CIConnector, CURRENT_SCHEMA_VERSION, type PackedEnvelope as CacheableEnvelope, type CascadeLayer, type CascadeNode, type CascadeResult, type CascadeSimulationOptions, CascadeSimulator, type ClassificationResult, CodeIngestor, CompositeProbabilityStrategy, type ConflictDetail, type ConflictPrediction, ConflictPredictor, type ConflictSeverity, type ConflictType, ConfluenceConnector, type ConnectorConfig, ContextQL, type ContextQLParams, type ContextQLResult, type Contradiction, ContradictionDetector, type ContradictionEntry, type ContradictionResult, type CoverageReport, CoverageScorer, D2Parser, DEFAULT_BLOCKLIST, DEFAULT_PATTERNS, DecisionIngestor, DesignConstraintAdapter, DesignIngestor, type DesignStrictness, type DesignViolation, type DetectedElement, type DiagramEntity, type DiagramFormatParser, type DiagramParseResult, DiagramParser, type DiagramRelationship, type DomainCoverage, type DomainCoverageScore, type DomainInferenceOptions, type DriftClassification, type DriftDetector, type DriftFinding, type DriftResult, EDGE_TYPES, type EdgeQuery, type EdgeType, EntityExtractor, EntityResolver, EnumConstantExtractor, type ExtractionCounts, type ExtractionRecord, ExtractionRunner, FigmaConnector, FusionLayer, type FusionResult, type GapEntry, type GapReport, GitIngestor, type GitRunner, GraphAnomalyAdapter, type GraphBudget, GraphComplexityAdapter, type GraphComplexityHotspot, type GraphComplexityResult, type GraphConnector, GraphConstraintAdapter, GraphCouplingAdapter, type GraphCouplingFileData, type GraphCouplingResult, type GraphCoverageReport, type GraphDeadCodeData, type GraphDependencyData, type GraphDriftData, type GraphEdge, GraphEdgeSchema, GraphEntropyAdapter, GraphFeedbackAdapter, type GraphFilterResult, type GraphHarnessCheckData, type GraphImpactData, type GraphLayerViolation, type GraphMetadata, type GraphNode, GraphNodeSchema, type GraphSnapshotSummary, type GraphStabilityTier, GraphStore, type HttpClient, INTENTS, ImageAnalysisExtractor, type ImageAnalysisExtractorOptions, type AnalysisProvider as ImageAnalysisProvider, type ImageAnalysisResult, type ImpactGroups, type IndependenceCheckParams, type IndependenceResult, type IngestResult, type Intent, IntentClassifier, JiraConnector, KnowledgeDocMaterializer, KnowledgeIngestor, type KnowledgePipelineOptions, type KnowledgePipelineResult, KnowledgePipelineRunner, type KnowledgeSnapshot, type KnowledgeSnapshotEntry, KnowledgeStagingAggregator, type Language, type LinkResult, type LoadGraphResult, type MaterializeOptions, type MaterializeResult, type MaterializedDoc, MermaidParser, MiroConnector, NODE_STABILITY, NODE_TYPES, type NodeCategory, type NodeQuery, type NodeType, OBSERVABILITY_TYPES, type OverlapDetail, PackedSummaryCache, type PairResult, PlantUmlParser, type ProbabilityStrategy, type ProjectionSpec, type RequirementCoverage, RequirementIngestor, type ResolvedEntity, ResponseFormatter, type SignalExtractor, type SkippedEntry, SlackConnector, type SourceLocation, type StagedEntry, type StatisticalOutlier, StructuralDriftDetector, SyncManager, type SyncMetadata, type TaskDefinition, TaskIndependenceAnalyzer, TestDescriptionExtractor, TopologicalLinker, type TraceabilityOptions, type TraceabilityResult, type TracedFile, VERSION, ValidationRuleExtractor, type VectorSearchResult, VectorStore, askGraph, classifyNodeCategory, createExtractionRunner, detectLanguage, groupNodesByImpact, inferDomain, linkToCode, loadGraph, normalizeIntent, project, queryTraceability, saveGraph };

package/dist/index.d.ts

@@ -512,6 +512,33 @@ declare class RequirementIngestor {
     private linkByKeywordOverlap;
 }
 
+/**
+ * Domain inference for graph nodes.
+ *
+ * Precedence (highest to lowest):
+ *   1. node.metadata.domain (explicit)
+ *   2. extraPatterns (config-provided)
+ *   3. DEFAULT_PATTERNS (built-in)
+ *   4. Generic first non-blocklisted path segment
+ *   5. KnowledgeLinker connector source (path-less facts)
+ *   6. 'unknown'
+ *
+ * Pattern format: 'prefix/<dir>'. Single-segment prefix only. <dir> captures
+ * the segment immediately after the prefix.
+ */
+interface DomainInferenceOptions {
+    /** Additional patterns beyond the built-in defaults. Format: 'prefix/<dir>'. */
+    extraPatterns?: readonly string[];
+    /** Additional blocklisted segments beyond the built-in defaults. */
+    extraBlocklist?: readonly string[];
+}
+declare const DEFAULT_PATTERNS: readonly string[];
+declare const DEFAULT_BLOCKLIST: ReadonlySet<string>;
+declare function inferDomain(node: {
+    path?: string;
+    metadata?: Record<string, unknown>;
+}, options?: DomainInferenceOptions): string;
+
 type DriftClassification = 'new' | 'drifted' | 'stale' | 'contradicting';
 interface KnowledgeSnapshotEntry {
     readonly id: string;
@@ -591,7 +618,8 @@ interface AggregateResult {
 }
 declare class KnowledgeStagingAggregator {
     private readonly projectDir;
-    constructor(projectDir: string);
+    private readonly inferenceOptions;
+    constructor(projectDir: string, inferenceOptions?: DomainInferenceOptions);
     aggregate(extractorResults: readonly StagedEntry[], linkerResults: readonly StagedEntry[], diagramResults: readonly StagedEntry[]): Promise<AggregateResult>;
     private extractDocName;
     generateGapReport(knowledgeDir: string, store?: GraphStore): Promise<GapReport>;
@@ -691,8 +719,9 @@ interface CoverageReport {
     readonly generatedAt: string;
 }
 declare class CoverageScorer {
+    private readonly inferenceOptions;
+    constructor(inferenceOptions?: DomainInferenceOptions);
     score(store: GraphStore): CoverageReport;
-    private domainFromPath;
 }
 
 /**
@@ -725,7 +754,8 @@ interface SkippedEntry {
 }
 declare class KnowledgeDocMaterializer {
     private readonly store;
-    constructor(store: GraphStore);
+    private readonly inferenceOptions;
+    constructor(store: GraphStore, inferenceOptions?: DomainInferenceOptions);
     materialize(gapEntries: readonly GapEntry[], options: MaterializeOptions): Promise<MaterializeResult>;
     private resolveEntry;
     inferDomain(node: GraphNode): string | null;
@@ -758,6 +788,13 @@ interface KnowledgePipelineOptions {
     readonly analyzeImages?: boolean;
     readonly analysisProvider?: AnalysisProvider;
     readonly imagePaths?: readonly string[];
+    /**
+     * Domain-inference overrides threaded into KnowledgeStagingAggregator,
+     * CoverageScorer, and KnowledgeDocMaterializer. Sourced by the CLI from
+     * `harness.config.json#knowledge.domainPatterns` (→ extraPatterns) and
+     * `knowledge.domainBlocklist` (→ extraBlocklist). Defaults to {} when absent.
+     */
+    readonly inferenceOptions?: DomainInferenceOptions;
 }
 interface ExtractionCounts {
     readonly codeSignals: number;
@@ -782,6 +819,8 @@ interface KnowledgePipelineResult {
 declare class KnowledgePipelineRunner {
     private readonly store;
     constructor(store: GraphStore);
+    /** Resolved per-`run()` inference options. Set on entry to `run()`. */
+    private inferenceOptions;
     run(options: KnowledgePipelineOptions): Promise<KnowledgePipelineResult>;
     /** Run the remediation convergence loop; returns iteration count and accumulated materialization. */
     private runRemediationLoop;
@@ -1827,4 +1866,4 @@ declare class CascadeSimulator {
 
 declare const VERSION = "0.6.0";
 
-export { ALL_EXTRACTORS, type AggregateResult, type AnomalyDetectionOptions, type AnomalyReport, ApiPathExtractor, type ArticulationPoint, type AskGraphResult, type AssembledContext, Assembler, BusinessKnowledgeIngestor, CIConnector, CURRENT_SCHEMA_VERSION, type PackedEnvelope as CacheableEnvelope, type CascadeLayer, type CascadeNode, type CascadeResult, type CascadeSimulationOptions, CascadeSimulator, type ClassificationResult, CodeIngestor, CompositeProbabilityStrategy, type ConflictDetail, type ConflictPrediction, ConflictPredictor, type ConflictSeverity, type ConflictType, ConfluenceConnector, type ConnectorConfig, ContextQL, type ContextQLParams, type ContextQLResult, type Contradiction, ContradictionDetector, type ContradictionEntry, type ContradictionResult, type CoverageReport, CoverageScorer, D2Parser, DecisionIngestor, DesignConstraintAdapter, DesignIngestor, type DesignStrictness, type DesignViolation, type DetectedElement, type DiagramEntity, type DiagramFormatParser, type DiagramParseResult, DiagramParser, type DiagramRelationship, type DomainCoverage, type DomainCoverageScore, type DriftClassification, type DriftDetector, type DriftFinding, type DriftResult, EDGE_TYPES, type EdgeQuery, type EdgeType, EntityExtractor, EntityResolver, EnumConstantExtractor, type ExtractionCounts, type ExtractionRecord, ExtractionRunner, FigmaConnector, FusionLayer, type FusionResult, type GapEntry, type GapReport, GitIngestor, type GitRunner, GraphAnomalyAdapter, type GraphBudget, GraphComplexityAdapter, type GraphComplexityHotspot, type GraphComplexityResult, type GraphConnector, GraphConstraintAdapter, GraphCouplingAdapter, type GraphCouplingFileData, type GraphCouplingResult, type GraphCoverageReport, type GraphDeadCodeData, type GraphDependencyData, type GraphDriftData, type GraphEdge, GraphEdgeSchema, GraphEntropyAdapter, GraphFeedbackAdapter, type GraphFilterResult, type GraphHarnessCheckData, type GraphImpactData, type GraphLayerViolation, type GraphMetadata, type GraphNode, GraphNodeSchema, type GraphSnapshotSummary, type GraphStabilityTier, GraphStore, type HttpClient, INTENTS, ImageAnalysisExtractor, type ImageAnalysisExtractorOptions, type AnalysisProvider as ImageAnalysisProvider, type ImageAnalysisResult, type ImpactGroups, type IndependenceCheckParams, type IndependenceResult, type IngestResult, type Intent, IntentClassifier, JiraConnector, KnowledgeDocMaterializer, KnowledgeIngestor, type KnowledgePipelineOptions, type KnowledgePipelineResult, KnowledgePipelineRunner, type KnowledgeSnapshot, type KnowledgeSnapshotEntry, KnowledgeStagingAggregator, type Language, type LinkResult, type LoadGraphResult, type MaterializeOptions, type MaterializeResult, type MaterializedDoc, MermaidParser, MiroConnector, NODE_STABILITY, NODE_TYPES, type NodeCategory, type NodeQuery, type NodeType, OBSERVABILITY_TYPES, type OverlapDetail, PackedSummaryCache, type PairResult, PlantUmlParser, type ProbabilityStrategy, type ProjectionSpec, type RequirementCoverage, RequirementIngestor, type ResolvedEntity, ResponseFormatter, type SignalExtractor, type SkippedEntry, SlackConnector, type SourceLocation, type StagedEntry, type StatisticalOutlier, StructuralDriftDetector, SyncManager, type SyncMetadata, type TaskDefinition, TaskIndependenceAnalyzer, TestDescriptionExtractor, TopologicalLinker, type TraceabilityOptions, type TraceabilityResult, type TracedFile, VERSION, ValidationRuleExtractor, type VectorSearchResult, VectorStore, askGraph, classifyNodeCategory, createExtractionRunner, detectLanguage, groupNodesByImpact, linkToCode, loadGraph, normalizeIntent, project, queryTraceability, saveGraph };
+export { ALL_EXTRACTORS, type AggregateResult, type AnomalyDetectionOptions, type AnomalyReport, ApiPathExtractor, type ArticulationPoint, type AskGraphResult, type AssembledContext, Assembler, BusinessKnowledgeIngestor, CIConnector, CURRENT_SCHEMA_VERSION, type PackedEnvelope as CacheableEnvelope, type CascadeLayer, type CascadeNode, type CascadeResult, type CascadeSimulationOptions, CascadeSimulator, type ClassificationResult, CodeIngestor, CompositeProbabilityStrategy, type ConflictDetail, type ConflictPrediction, ConflictPredictor, type ConflictSeverity, type ConflictType, ConfluenceConnector, type ConnectorConfig, ContextQL, type ContextQLParams, type ContextQLResult, type Contradiction, ContradictionDetector, type ContradictionEntry, type ContradictionResult, type CoverageReport, CoverageScorer, D2Parser, DEFAULT_BLOCKLIST, DEFAULT_PATTERNS, DecisionIngestor, DesignConstraintAdapter, DesignIngestor, type DesignStrictness, type DesignViolation, type DetectedElement, type DiagramEntity, type DiagramFormatParser, type DiagramParseResult, DiagramParser, type DiagramRelationship, type DomainCoverage, type DomainCoverageScore, type DomainInferenceOptions, type DriftClassification, type DriftDetector, type DriftFinding, type DriftResult, EDGE_TYPES, type EdgeQuery, type EdgeType, EntityExtractor, EntityResolver, EnumConstantExtractor, type ExtractionCounts, type ExtractionRecord, ExtractionRunner, FigmaConnector, FusionLayer, type FusionResult, type GapEntry, type GapReport, GitIngestor, type GitRunner, GraphAnomalyAdapter, type GraphBudget, GraphComplexityAdapter, type GraphComplexityHotspot, type GraphComplexityResult, type GraphConnector, GraphConstraintAdapter, GraphCouplingAdapter, type GraphCouplingFileData, type GraphCouplingResult, type GraphCoverageReport, type GraphDeadCodeData, type GraphDependencyData, type GraphDriftData, type GraphEdge, GraphEdgeSchema, GraphEntropyAdapter, GraphFeedbackAdapter, type GraphFilterResult, type GraphHarnessCheckData, type GraphImpactData, type GraphLayerViolation, type GraphMetadata, type GraphNode, GraphNodeSchema, type GraphSnapshotSummary, type GraphStabilityTier, GraphStore, type HttpClient, INTENTS, ImageAnalysisExtractor, type ImageAnalysisExtractorOptions, type AnalysisProvider as ImageAnalysisProvider, type ImageAnalysisResult, type ImpactGroups, type IndependenceCheckParams, type IndependenceResult, type IngestResult, type Intent, IntentClassifier, JiraConnector, KnowledgeDocMaterializer, KnowledgeIngestor, type KnowledgePipelineOptions, type KnowledgePipelineResult, KnowledgePipelineRunner, type KnowledgeSnapshot, type KnowledgeSnapshotEntry, KnowledgeStagingAggregator, type Language, type LinkResult, type LoadGraphResult, type MaterializeOptions, type MaterializeResult, type MaterializedDoc, MermaidParser, MiroConnector, NODE_STABILITY, NODE_TYPES, type NodeCategory, type NodeQuery, type NodeType, OBSERVABILITY_TYPES, type OverlapDetail, PackedSummaryCache, type PairResult, PlantUmlParser, type ProbabilityStrategy, type ProjectionSpec, type RequirementCoverage, RequirementIngestor, type ResolvedEntity, ResponseFormatter, type SignalExtractor, type SkippedEntry, SlackConnector, type SourceLocation, type StagedEntry, type StatisticalOutlier, StructuralDriftDetector, SyncManager, type SyncMetadata, type TaskDefinition, TaskIndependenceAnalyzer, TestDescriptionExtractor, TopologicalLinker, type TraceabilityOptions, type TraceabilityResult, type TracedFile, VERSION, ValidationRuleExtractor, type VectorSearchResult, VectorStore, askGraph, classifyNodeCategory, createExtractionRunner, detectLanguage, groupNodesByImpact, inferDomain, linkToCode, loadGraph, normalizeIntent, project, queryTraceability, saveGraph };

package/dist/index.js

@@ -45,6 +45,8 @@ __export(index_exports, {
   ContradictionDetector: () => ContradictionDetector,
   CoverageScorer: () => CoverageScorer,
   D2Parser: () => D2Parser,
+  DEFAULT_BLOCKLIST: () => DEFAULT_BLOCKLIST,
+  DEFAULT_PATTERNS: () => DEFAULT_PATTERNS,
   DecisionIngestor: () => DecisionIngestor,
   DesignConstraintAdapter: () => DesignConstraintAdapter,
   DesignIngestor: () => DesignIngestor,
@@ -97,6 +99,7 @@ __export(index_exports, {
   createExtractionRunner: () => createExtractionRunner,
   detectLanguage: () => detectLanguage,
   groupNodesByImpact: () => groupNodesByImpact,
+  inferDomain: () => inferDomain,
   linkToCode: () => linkToCode,
   loadGraph: () => loadGraph,
   normalizeIntent: () => normalizeIntent,
@@ -2496,6 +2499,88 @@ var RequirementIngestor = class {
   }
 };
 
+// src/ingest/domain-inference.ts
+var DEFAULT_PATTERNS = [
+  "packages/<dir>",
+  "apps/<dir>",
+  "services/<dir>",
+  "src/<dir>",
+  "lib/<dir>"
+];
+var DEFAULT_BLOCKLIST = /* @__PURE__ */ new Set([
+  "node_modules",
+  ".harness",
+  "dist",
+  "build",
+  ".git",
+  "coverage",
+  ".next",
+  ".turbo",
+  ".cache",
+  "out",
+  "tmp"
+]);
+var CODE_EXTENSIONS = [".ts", ".tsx", ".js", ".jsx", ".mjs", ".cjs"];
+function matchPattern(filePath, pattern) {
+  const patternParts = pattern.split("/").filter((s) => s.length > 0);
+  if (patternParts.length !== 2 || patternParts[1] !== "<dir>") {
+    return null;
+  }
+  const prefix = patternParts[0];
+  const pathParts = filePath.split("/").filter((s) => s.length > 0);
+  if (pathParts.length < 2) return null;
+  if (pathParts[0] !== prefix) return null;
+  let dir = pathParts[1];
+  if (dir.length === 0) return null;
+  if (CODE_EXTENSIONS.some((ext) => dir.endsWith(ext))) {
+    const dotIdx = dir.lastIndexOf(".");
+    if (dotIdx > 0) dir = dir.slice(0, dotIdx);
+  }
+  if (dir.length === 0) return null;
+  return dir;
+}
+function inferDomain(node, options = {}) {
+  if (node.metadata?.domain && typeof node.metadata.domain === "string" && node.metadata.domain.length > 0) {
+    return node.metadata.domain;
+  }
+  const filePath = typeof node.path === "string" ? node.path : "";
+  const blocklist = new Set(DEFAULT_BLOCKLIST);
+  if (options.extraBlocklist) {
+    for (const seg of options.extraBlocklist) {
+      if (seg && seg.length > 0) blocklist.add(seg);
+    }
+  }
+  if (filePath.length > 0) {
+    const extraPatterns = options.extraPatterns ?? [];
+    for (const pattern of extraPatterns) {
+      const dir = matchPattern(filePath, pattern);
+      if (dir !== null) {
+        if (blocklist.has(dir)) return "unknown";
+        return dir;
+      }
+    }
+    for (const pattern of DEFAULT_PATTERNS) {
+      const dir = matchPattern(filePath, pattern);
+      if (dir !== null) {
+        if (blocklist.has(dir)) return "unknown";
+        return dir;
+      }
+    }
+    const segments = filePath.split("/").filter((s) => s.length > 0);
+    if (segments.length > 0) {
+      const first = segments[0];
+      if (!blocklist.has(first)) return first;
+    }
+  }
+  const source = node.metadata?.source;
+  if (source === "knowledge-linker" || source === "connector") {
+    const connector = node.metadata?.connectorName;
+    if (typeof connector === "string" && connector.length > 0) return connector;
+    return "general";
+  }
+  return "unknown";
+}
+
 // src/ingest/KnowledgePipelineRunner.ts
 var fs11 = __toESM(require("fs/promises"));
 var path12 = __toESM(require("path"));
@@ -3308,10 +3393,12 @@ var BUSINESS_NODE_TYPES = [
   "business_fact"
 ];
 var KnowledgeStagingAggregator = class {
-  constructor(projectDir) {
+  constructor(projectDir, inferenceOptions = {}) {
     this.projectDir = projectDir;
+    this.inferenceOptions = inferenceOptions;
   }
   projectDir;
+  inferenceOptions;
   async aggregate(extractorResults, linkerResults, diagramResults) {
     const all = [...extractorResults, ...linkerResults, ...diagramResults];
     const byHash = /* @__PURE__ */ new Map();
@@ -3380,7 +3467,7 @@ var KnowledgeStagingAggregator = class {
     for (const nodeType of BUSINESS_NODE_TYPES) {
       const nodes = store.findNodes({ type: nodeType });
       for (const node of nodes) {
-        const domain = node.metadata?.domain ?? "unknown";
+        const domain = inferDomain(node, this.inferenceOptions);
         const list = extractedByDomain.get(domain) ?? [];
         list.push(node);
         extractedByDomain.set(domain, list);
@@ -5058,10 +5145,10 @@ function toGrade(score) {
   if (score >= 20) return "D";
   return "F";
 }
-function groupByDomain(nodes, fallback) {
+function groupByDomain(nodes, _fallback, options = {}) {
   const map = /* @__PURE__ */ new Map();
   for (const node of nodes) {
-    const domain = node.metadata.domain ?? fallback(node);
+    const domain = inferDomain(node, options);
     const group = map.get(domain) ?? [];
     group.push(node);
     map.set(domain, group);
@@ -5117,11 +5204,15 @@ function scoreDomain(domain, knEntries, codeEntries, store) {
   };
 }
 var CoverageScorer = class {
+  constructor(inferenceOptions = {}) {
+    this.inferenceOptions = inferenceOptions;
+  }
+  inferenceOptions;
   score(store) {
     const knowledgeNodes = KNOWLEDGE_TYPES.flatMap((t) => store.findNodes({ type: t }));
-    const domainMap = groupByDomain(knowledgeNodes, () => "unclassified");
+    const domainMap = groupByDomain(knowledgeNodes, void 0, this.inferenceOptions);
     const codeNodes = CODE_TYPES2.flatMap((t) => store.findNodes({ type: t }));
-    const codeDomains = groupByDomain(codeNodes, (n) => this.domainFromPath(n.path));
+    const codeDomains = groupByDomain(codeNodes, void 0, this.inferenceOptions);
     const allDomains = /* @__PURE__ */ new Set([...domainMap.keys(), ...codeDomains.keys()]);
     const domains = [];
     for (const domain of allDomains) {
@@ -5137,15 +5228,6 @@ var CoverageScorer = class {
       generatedAt: (/* @__PURE__ */ new Date()).toISOString()
     };
   }
-  domainFromPath(filePath) {
-    if (!filePath) return "unclassified";
-    const parts = filePath.split("/");
-    const pkgIdx = parts.indexOf("packages");
-    if (pkgIdx >= 0 && parts[pkgIdx + 1]) return parts[pkgIdx + 1];
-    const srcIdx = parts.indexOf("src");
-    if (srcIdx >= 0 && parts[srcIdx + 1]) return parts[srcIdx + 1];
-    return parts[0] ?? "unclassified";
-  }
 };
 
 // src/ingest/KnowledgeDocMaterializer.ts
@@ -5161,10 +5243,12 @@ var VALID_BUSINESS_TYPES = /* @__PURE__ */ new Set([
 var DEFAULT_MAX_DOCS = 50;
 var MAX_COLLISION_SUFFIX = 10;
 var KnowledgeDocMaterializer = class {
-  constructor(store) {
+  constructor(store, inferenceOptions = {}) {
     this.store = store;
+    this.inferenceOptions = inferenceOptions;
   }
   store;
+  inferenceOptions;
   async materialize(gapEntries, options) {
     const maxDocs = options.maxDocs ?? DEFAULT_MAX_DOCS;
     const created = [];
@@ -5220,21 +5304,8 @@ var KnowledgeDocMaterializer = class {
     return { node, domain, domainDir, nameKey };
   }
   inferDomain(node) {
-    if (node.metadata?.domain && typeof node.metadata.domain === "string") {
-      return node.metadata.domain;
-    }
-    if (node.path) {
-      const pkgMatch = node.path.match(/^packages\/([^/]+)/);
-      if (pkgMatch) return pkgMatch[1];
-      const srcMatch = node.path.match(/^src\/([^/]+)/);
-      if (srcMatch) return srcMatch[1];
-    }
-    if (node.metadata?.source === "knowledge-linker" || node.metadata?.source === "connector") {
-      const connector = node.metadata.connectorName;
-      if (typeof connector === "string") return connector;
-      return "general";
-    }
-    return null;
+    const result = inferDomain(node, this.inferenceOptions);
+    return result === "unknown" ? null : result;
   }
   generateFilename(name) {
     const slug = name.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-|-$/g, "").slice(0, 60);
@@ -5328,7 +5399,10 @@ var KnowledgePipelineRunner = class {
     this.store = store;
   }
   store;
+  /** Resolved per-`run()` inference options. Set on entry to `run()`. */
+  inferenceOptions = {};
   async run(options) {
+    this.inferenceOptions = options.inferenceOptions ?? {};
     const remediations = [];
     const preSnapshot = this.buildSnapshot(options.domain);
     const extraction = await this.extract(options);
@@ -5336,7 +5410,7 @@ var KnowledgePipelineRunner = class {
     let driftResult = this.reconcile(preSnapshot, postSnapshot);
     const contradictions = new ContradictionDetector().detect(this.store);
     let gapReport = await this.detect(options);
-    const coverage = new CoverageScorer().score(this.store);
+    const coverage = new CoverageScorer(this.inferenceOptions).score(this.store);
     let materialization;
     let iterations = 1;
     if (options.fix) {
@@ -5499,7 +5573,7 @@ var KnowledgePipelineRunner = class {
   // ── Phase 3: DETECT ───────────────────────────────────────────────────────
   async detect(options) {
     const knowledgeDir = path12.join(options.projectDir, "docs", "knowledge");
-    const aggregator = new KnowledgeStagingAggregator(options.projectDir);
+    const aggregator = new KnowledgeStagingAggregator(options.projectDir, this.inferenceOptions);
     const gapReport = await aggregator.generateGapReport(knowledgeDir, this.store);
     await aggregator.writeGapReport(gapReport);
     return gapReport;
@@ -5527,7 +5601,7 @@ var KnowledgePipelineRunner = class {
       const allGapEntries = gapReport.domains.flatMap((d) => d.gapEntries);
       const materializable = allGapEntries.filter((e) => e.hasContent);
       if (materializable.length > 0) {
-        const materializer = new KnowledgeDocMaterializer(this.store);
+        const materializer = new KnowledgeDocMaterializer(this.store, this.inferenceOptions);
         const matResult = await materializer.materialize(materializable, {
           projectDir: options.projectDir,
           dryRun: false
@@ -5553,7 +5627,7 @@ var KnowledgePipelineRunner = class {
       timestamp: (/* @__PURE__ */ new Date()).toISOString()
     }));
     if (stagedEntries.length > 0) {
-      const aggregator = new KnowledgeStagingAggregator(options.projectDir);
+      const aggregator = new KnowledgeStagingAggregator(options.projectDir, this.inferenceOptions);
       await aggregator.aggregate(stagedEntries, [], []);
     }
   }
@@ -9527,6 +9601,8 @@ var VERSION = "0.6.0";
   ContradictionDetector,
   CoverageScorer,
   D2Parser,
+  DEFAULT_BLOCKLIST,
+  DEFAULT_PATTERNS,
   DecisionIngestor,
   DesignConstraintAdapter,
   DesignIngestor,
@@ -9579,6 +9655,7 @@ var VERSION = "0.6.0";
   createExtractionRunner,
   detectLanguage,
   groupNodesByImpact,
+  inferDomain,
   linkToCode,
   loadGraph,
   normalizeIntent,

package/dist/index.mjs

@@ -2388,6 +2388,88 @@ var RequirementIngestor = class {
   }
 };
 
+// src/ingest/domain-inference.ts
+var DEFAULT_PATTERNS = [
+  "packages/<dir>",
+  "apps/<dir>",
+  "services/<dir>",
+  "src/<dir>",
+  "lib/<dir>"
+];
+var DEFAULT_BLOCKLIST = /* @__PURE__ */ new Set([
+  "node_modules",
+  ".harness",
+  "dist",
+  "build",
+  ".git",
+  "coverage",
+  ".next",
+  ".turbo",
+  ".cache",
+  "out",
+  "tmp"
+]);
+var CODE_EXTENSIONS = [".ts", ".tsx", ".js", ".jsx", ".mjs", ".cjs"];
+function matchPattern(filePath, pattern) {
+  const patternParts = pattern.split("/").filter((s) => s.length > 0);
+  if (patternParts.length !== 2 || patternParts[1] !== "<dir>") {
+    return null;
+  }
+  const prefix = patternParts[0];
+  const pathParts = filePath.split("/").filter((s) => s.length > 0);
+  if (pathParts.length < 2) return null;
+  if (pathParts[0] !== prefix) return null;
+  let dir = pathParts[1];
+  if (dir.length === 0) return null;
+  if (CODE_EXTENSIONS.some((ext) => dir.endsWith(ext))) {
+    const dotIdx = dir.lastIndexOf(".");
+    if (dotIdx > 0) dir = dir.slice(0, dotIdx);
+  }
+  if (dir.length === 0) return null;
+  return dir;
+}
+function inferDomain(node, options = {}) {
+  if (node.metadata?.domain && typeof node.metadata.domain === "string" && node.metadata.domain.length > 0) {
+    return node.metadata.domain;
+  }
+  const filePath = typeof node.path === "string" ? node.path : "";
+  const blocklist = new Set(DEFAULT_BLOCKLIST);
+  if (options.extraBlocklist) {
+    for (const seg of options.extraBlocklist) {
+      if (seg && seg.length > 0) blocklist.add(seg);
+    }
+  }
+  if (filePath.length > 0) {
+    const extraPatterns = options.extraPatterns ?? [];
+    for (const pattern of extraPatterns) {
+      const dir = matchPattern(filePath, pattern);
+      if (dir !== null) {
+        if (blocklist.has(dir)) return "unknown";
+        return dir;
+      }
+    }
+    for (const pattern of DEFAULT_PATTERNS) {
+      const dir = matchPattern(filePath, pattern);
+      if (dir !== null) {
+        if (blocklist.has(dir)) return "unknown";
+        return dir;
+      }
+    }
+    const segments = filePath.split("/").filter((s) => s.length > 0);
+    if (segments.length > 0) {
+      const first = segments[0];
+      if (!blocklist.has(first)) return first;
+    }
+  }
+  const source = node.metadata?.source;
+  if (source === "knowledge-linker" || source === "connector") {
+    const connector = node.metadata?.connectorName;
+    if (typeof connector === "string" && connector.length > 0) return connector;
+    return "general";
+  }
+  return "unknown";
+}
+
 // src/ingest/KnowledgePipelineRunner.ts
 import * as fs11 from "fs/promises";
 import * as path12 from "path";
@@ -3200,10 +3282,12 @@ var BUSINESS_NODE_TYPES = [
   "business_fact"
 ];
 var KnowledgeStagingAggregator = class {
-  constructor(projectDir) {
+  constructor(projectDir, inferenceOptions = {}) {
     this.projectDir = projectDir;
+    this.inferenceOptions = inferenceOptions;
   }
   projectDir;
+  inferenceOptions;
   async aggregate(extractorResults, linkerResults, diagramResults) {
     const all = [...extractorResults, ...linkerResults, ...diagramResults];
     const byHash = /* @__PURE__ */ new Map();
@@ -3272,7 +3356,7 @@ var KnowledgeStagingAggregator = class {
     for (const nodeType of BUSINESS_NODE_TYPES) {
       const nodes = store.findNodes({ type: nodeType });
       for (const node of nodes) {
-        const domain = node.metadata?.domain ?? "unknown";
+        const domain = inferDomain(node, this.inferenceOptions);
         const list = extractedByDomain.get(domain) ?? [];
         list.push(node);
         extractedByDomain.set(domain, list);
@@ -4950,10 +5034,10 @@ function toGrade(score) {
   if (score >= 20) return "D";
   return "F";
 }
-function groupByDomain(nodes, fallback) {
+function groupByDomain(nodes, _fallback, options = {}) {
   const map = /* @__PURE__ */ new Map();
   for (const node of nodes) {
-    const domain = node.metadata.domain ?? fallback(node);
+    const domain = inferDomain(node, options);
     const group = map.get(domain) ?? [];
     group.push(node);
     map.set(domain, group);
@@ -5009,11 +5093,15 @@ function scoreDomain(domain, knEntries, codeEntries, store) {
   };
 }
 var CoverageScorer = class {
+  constructor(inferenceOptions = {}) {
+    this.inferenceOptions = inferenceOptions;
+  }
+  inferenceOptions;
   score(store) {
     const knowledgeNodes = KNOWLEDGE_TYPES.flatMap((t) => store.findNodes({ type: t }));
-    const domainMap = groupByDomain(knowledgeNodes, () => "unclassified");
+    const domainMap = groupByDomain(knowledgeNodes, void 0, this.inferenceOptions);
     const codeNodes = CODE_TYPES2.flatMap((t) => store.findNodes({ type: t }));
-    const codeDomains = groupByDomain(codeNodes, (n) => this.domainFromPath(n.path));
+    const codeDomains = groupByDomain(codeNodes, void 0, this.inferenceOptions);
     const allDomains = /* @__PURE__ */ new Set([...domainMap.keys(), ...codeDomains.keys()]);
     const domains = [];
     for (const domain of allDomains) {
@@ -5029,15 +5117,6 @@ var CoverageScorer = class {
       generatedAt: (/* @__PURE__ */ new Date()).toISOString()
     };
   }
-  domainFromPath(filePath) {
-    if (!filePath) return "unclassified";
-    const parts = filePath.split("/");
-    const pkgIdx = parts.indexOf("packages");
-    if (pkgIdx >= 0 && parts[pkgIdx + 1]) return parts[pkgIdx + 1];
-    const srcIdx = parts.indexOf("src");
-    if (srcIdx >= 0 && parts[srcIdx + 1]) return parts[srcIdx + 1];
-    return parts[0] ?? "unclassified";
-  }
 };
 
 // src/ingest/KnowledgeDocMaterializer.ts
@@ -5053,10 +5132,12 @@ var VALID_BUSINESS_TYPES = /* @__PURE__ */ new Set([
 var DEFAULT_MAX_DOCS = 50;
 var MAX_COLLISION_SUFFIX = 10;
 var KnowledgeDocMaterializer = class {
-  constructor(store) {
+  constructor(store, inferenceOptions = {}) {
     this.store = store;
+    this.inferenceOptions = inferenceOptions;
   }
   store;
+  inferenceOptions;
   async materialize(gapEntries, options) {
     const maxDocs = options.maxDocs ?? DEFAULT_MAX_DOCS;
     const created = [];
@@ -5112,21 +5193,8 @@ var KnowledgeDocMaterializer = class {
     return { node, domain, domainDir, nameKey };
   }
   inferDomain(node) {
-    if (node.metadata?.domain && typeof node.metadata.domain === "string") {
-      return node.metadata.domain;
-    }
-    if (node.path) {
-      const pkgMatch = node.path.match(/^packages\/([^/]+)/);
-      if (pkgMatch) return pkgMatch[1];
-      const srcMatch = node.path.match(/^src\/([^/]+)/);
-      if (srcMatch) return srcMatch[1];
-    }
-    if (node.metadata?.source === "knowledge-linker" || node.metadata?.source === "connector") {
-      const connector = node.metadata.connectorName;
-      if (typeof connector === "string") return connector;
-      return "general";
-    }
-    return null;
+    const result = inferDomain(node, this.inferenceOptions);
+    return result === "unknown" ? null : result;
   }
   generateFilename(name) {
     const slug = name.toLowerCase().replace(/[^a-z0-9]+/g, "-").replace(/^-|-$/g, "").slice(0, 60);
@@ -5220,7 +5288,10 @@ var KnowledgePipelineRunner = class {
     this.store = store;
   }
   store;
+  /** Resolved per-`run()` inference options. Set on entry to `run()`. */
+  inferenceOptions = {};
   async run(options) {
+    this.inferenceOptions = options.inferenceOptions ?? {};
     const remediations = [];
     const preSnapshot = this.buildSnapshot(options.domain);
     const extraction = await this.extract(options);
@@ -5228,7 +5299,7 @@ var KnowledgePipelineRunner = class {
     let driftResult = this.reconcile(preSnapshot, postSnapshot);
     const contradictions = new ContradictionDetector().detect(this.store);
     let gapReport = await this.detect(options);
-    const coverage = new CoverageScorer().score(this.store);
+    const coverage = new CoverageScorer(this.inferenceOptions).score(this.store);
     let materialization;
     let iterations = 1;
     if (options.fix) {
@@ -5391,7 +5462,7 @@ var KnowledgePipelineRunner = class {
   // ── Phase 3: DETECT ───────────────────────────────────────────────────────
   async detect(options) {
     const knowledgeDir = path12.join(options.projectDir, "docs", "knowledge");
-    const aggregator = new KnowledgeStagingAggregator(options.projectDir);
+    const aggregator = new KnowledgeStagingAggregator(options.projectDir, this.inferenceOptions);
     const gapReport = await aggregator.generateGapReport(knowledgeDir, this.store);
     await aggregator.writeGapReport(gapReport);
     return gapReport;
@@ -5419,7 +5490,7 @@ var KnowledgePipelineRunner = class {
       const allGapEntries = gapReport.domains.flatMap((d) => d.gapEntries);
       const materializable = allGapEntries.filter((e) => e.hasContent);
       if (materializable.length > 0) {
-        const materializer = new KnowledgeDocMaterializer(this.store);
+        const materializer = new KnowledgeDocMaterializer(this.store, this.inferenceOptions);
         const matResult = await materializer.materialize(materializable, {
           projectDir: options.projectDir,
           dryRun: false
@@ -5445,7 +5516,7 @@ var KnowledgePipelineRunner = class {
       timestamp: (/* @__PURE__ */ new Date()).toISOString()
     }));
     if (stagedEntries.length > 0) {
-      const aggregator = new KnowledgeStagingAggregator(options.projectDir);
+      const aggregator = new KnowledgeStagingAggregator(options.projectDir, this.inferenceOptions);
       await aggregator.aggregate(stagedEntries, [], []);
     }
   }
@@ -9418,6 +9489,8 @@ export {
   ContradictionDetector,
   CoverageScorer,
   D2Parser,
+  DEFAULT_BLOCKLIST,
+  DEFAULT_PATTERNS,
   DecisionIngestor,
   DesignConstraintAdapter,
   DesignIngestor,
@@ -9470,6 +9543,7 @@ export {
   createExtractionRunner,
   detectLanguage,
   groupNodesByImpact,
+  inferDomain,
   linkToCode,
   loadGraph,
   normalizeIntent,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@harness-engineering/graph",
-  "version": "0.6.0",
+  "version": "0.7.0",
   "license": "MIT",
   "description": "Knowledge graph for context assembly in Harness Engineering",
   "main": "./dist/index.js",