@fortemi/core 2026.7.1 → 2026.7.2

package/dist/index.d.ts

@@ -1,6 +1,6 @@
 import { PGlite } from '@electric-sql/pglite';
 import { z, ZodType } from 'zod';
-export { AIWG_SCAN_REQUIRED_FIELDS, AiwgChunkedIndexBuildOptions, AiwgChunkedIndexBuildResult, AiwgChunkedIndexDetailLoader, AiwgChunkedIndexLoadOptions, AiwgChunkedIndexLoader, AiwgChunkedIndexProgress, AiwgChunkedIndexProgressPhase, AiwgChunkedIndexQueryOptions, AiwgChunkedIndexQueryResult, AiwgChunkedIndexValidationResult, AiwgFortemiChunk, AiwgFortemiChunkDetailRef, AiwgFortemiChunkManifest, AiwgFortemiChunkPart, AiwgFortemiChunkPartRef, AiwgFortemiIndexExport, AiwgFortemiIndexExportSchemaVersion, AiwgFortemiKnownRecordType, AiwgFortemiProjectedRecord, AiwgFortemiProvenance, AiwgFortemiProvenanceEvent, AiwgFortemiRecord, AiwgFortemiRecordEmbedding, AiwgFortemiRecordSchemaVersion, AiwgFortemiRecordSource, AiwgFortemiRecordType, AiwgFortemiRelationship, AiwgFortemiRelationshipDirection, AiwgFortemiSearchProjection, AiwgFortemiSkosConcept, AiwgFortemiSkosRelation, AiwgFortemiSkosRelationType, AiwgIndexController, AiwgIndexControllerListener, AiwgIndexControllerSnapshot, AiwgIndexGraphOptions, AiwgIndexQueryMatch, AiwgIndexQueryOptions, AiwgIndexQueryRankedItem, AiwgIndexQueryResult, AiwgIndexQueryWeights, AiwgIndexValidationResult, AiwgPrivacyClassification, AiwgProvenanceConfidence, AiwgRelationshipDirection, AiwgRelationshipEdgeSummary, AiwgRelationshipNodeSummary, AiwgRelationshipQueryOptions, AiwgRelationshipSetOperation, AiwgRelationshipSetOptions, AiwgRelationshipSetResult, AiwgRelationshipTraversalOptions, AiwgRelationshipTraversalResult, AiwgReviewAction, AiwgReviewDecision, AiwgReviewDecisionExport, AiwgReviewInput, AiwgStaticDuplicatePair, AiwgStaticEmbeddingRecord, AiwgStaticEmbeddingSet, AiwgStaticHybridQueryOptions, AiwgStaticSemanticQueryOptions, AiwgStaticSemanticResult, aiwgFortemiIndexToCommunityGraph, assertAiwgFortemiChunkManifest, assertAiwgFortemiChunkPart, assertAiwgFortemiIndexExport, assertAiwgStaticEmbeddingSet, buildAiwgChunkedIndex, createAiwgFetchChunkLoader, createAiwgFetchDetailLoader, createAiwgIndexController, createAiwgReviewDecisionExport, findAiwgStaticDuplicatePairs, getAiwgFortemiFacets, queryAiwgFortemiIndex, queryAiwgHybridIndex, queryAiwgSemanticIndex, validateAiwgFortemiChunkManifest, validateAiwgFortemiChunkPart, validateAiwgFortemiIndexExport, validateAiwgStaticEmbeddingSet } from './aiwg-index.js';
+export { AIWG_SCAN_REQUIRED_FIELDS, AiwgChunkedIndexBuildOptions, AiwgChunkedIndexBuildResult, AiwgChunkedIndexDetailLoader, AiwgChunkedIndexLoadOptions, AiwgChunkedIndexLoader, AiwgChunkedIndexProgress, AiwgChunkedIndexProgressPhase, AiwgChunkedIndexQueryOptions, AiwgChunkedIndexQueryResult, AiwgChunkedIndexValidationResult, AiwgFortemiAttachmentReference, AiwgFortemiBinarySource, AiwgFortemiChunk, AiwgFortemiChunkDetailRef, AiwgFortemiChunkManifest, AiwgFortemiChunkPart, AiwgFortemiChunkPartRef, AiwgFortemiIndexExport, AiwgFortemiIndexExportSchemaVersion, AiwgFortemiKnownRecordType, AiwgFortemiProjectedRecord, AiwgFortemiProvenance, AiwgFortemiProvenanceEvent, AiwgFortemiRecord, AiwgFortemiRecordEmbedding, AiwgFortemiRecordSchemaVersion, AiwgFortemiRecordSource, AiwgFortemiRecordType, AiwgFortemiRelationship, AiwgFortemiRelationshipDirection, AiwgFortemiSearchProjection, AiwgFortemiSkosConcept, AiwgFortemiSkosRelation, AiwgFortemiSkosRelationType, AiwgHeadlessEmbeddingBackend, AiwgIndexController, AiwgIndexControllerListener, AiwgIndexControllerSnapshot, AiwgIndexGraphOptions, AiwgIndexQueryMatch, AiwgIndexQueryOptions, AiwgIndexQueryRankedItem, AiwgIndexQueryResult, AiwgIndexQueryWeights, AiwgIndexValidationResult, AiwgPrivacyClassification, AiwgProvenanceConfidence, AiwgRelationshipDirection, AiwgRelationshipEdgeSummary, AiwgRelationshipNodeSummary, AiwgRelationshipQueryOptions, AiwgRelationshipSetOperation, AiwgRelationshipSetOptions, AiwgRelationshipSetResult, AiwgRelationshipTraversalOptions, AiwgRelationshipTraversalResult, AiwgReviewAction, AiwgReviewDecision, AiwgReviewDecisionExport, AiwgReviewInput, AiwgStaticDuplicatePair, AiwgStaticEmbeddingRecord, AiwgStaticEmbeddingSet, AiwgStaticHybridQueryOptions, AiwgStaticSemanticQueryOptions, AiwgStaticSemanticResult, BuildAiwgStaticEmbeddingSetOptions, aiwgFortemiIndexToCommunityGraph, assertAiwgFortemiChunkManifest, assertAiwgFortemiChunkPart, assertAiwgFortemiIndexExport, assertAiwgStaticEmbeddingSet, buildAiwgChunkedIndex, buildAiwgStaticEmbeddingSet, createAiwgFetchChunkLoader, createAiwgFetchDetailLoader, createAiwgIndexController, createAiwgReviewDecisionExport, findAiwgStaticDuplicatePairs, getAiwgFortemiFacets, queryAiwgFortemiIndex, queryAiwgHybridIndex, queryAiwgSemanticIndex, validateAiwgFortemiChunkManifest, validateAiwgFortemiChunkPart, validateAiwgFortemiIndexExport, validateAiwgStaticEmbeddingSet } from './aiwg-index.js';
 
 /**
  * Generate a RFC 9562 UUIDv7 identifier.
@@ -386,6 +386,17 @@ declare const CURRENT_SHARD_VERSION = "1.0.0";
 declare const SHARD_FORMAT = "matric-shard";
 /** Components that can appear in a shard archive. */
 type ShardComponent = 'notes' | 'collections' | 'tags' | 'links' | 'embedding_sets' | 'embedding_set_members' | 'embedding_configs' | 'embeddings' | 'skos_schemes' | 'skos_concepts' | 'skos_relations' | 'note_skos_tags' | 'provenance_edges' | 'community_assignments' | 'communities' | 'graph_edges' | 'graph_sources';
+interface ShardAttachmentReference {
+    id: string;
+    path: string;
+    mime: string | null;
+    checksum: string;
+    bytes: number;
+}
+interface ShardBinarySource {
+    extracted_text: string;
+    attachment: ShardAttachmentReference;
+}
 /**
  * Reference to one cluster file of a component split across addressable files
  * (`notes/000.jsonl`, `notes/001.jsonl`, …). `offset`/`count` are the record
@@ -489,6 +500,7 @@ interface ShardNote {
     title: string | null;
     original_content: string;
     revised_content: string | null;
+    binary_sources?: ShardBinarySource[];
     format: string;
     source: string;
     starred: boolean;
@@ -705,6 +717,7 @@ interface BrowserNoteExport {
     deleted_at: Date | string | null;
     original_content: string;
     revised_content: string | null;
+    binary_sources?: ShardBinarySource[];
     tags: string[];
 }
 /** Convert a browser note to shard format. */
@@ -2733,6 +2746,8 @@ interface AttachmentRow {
     note_id: string;
     blob_id: string;
     document_type_id: string | null;
+    mime_type: string | null;
+    extracted_text: string | null;
     filename: string;
     display_name: string | null;
     position: number;
@@ -2751,6 +2766,7 @@ interface AttachInput {
     data: Uint8Array;
     filename: string;
     mimeType?: string;
+    extractedText?: string;
     displayName?: string;
 }
 declare class AttachmentsRepository {
@@ -2797,11 +2813,13 @@ declare const ManageAttachmentsInputSchema: z.ZodObject<{
     data_base64: z.ZodOptional<z.ZodString>;
     filename: z.ZodOptional<z.ZodString>;
     mime_type: z.ZodOptional<z.ZodString>;
+    extracted_text: z.ZodOptional<z.ZodString>;
     display_name: z.ZodOptional<z.ZodString>;
 }, "strip", z.ZodTypeAny, {
     action: "delete" | "list" | "attach" | "get" | "get_blob";
     note_id?: string | undefined;
     filename?: string | undefined;
+    extracted_text?: string | undefined;
     attachment_id?: string | undefined;
     data_base64?: string | undefined;
     mime_type?: string | undefined;
@@ -2810,6 +2828,7 @@ declare const ManageAttachmentsInputSchema: z.ZodObject<{
     action: "delete" | "list" | "attach" | "get" | "get_blob";
     note_id?: string | undefined;
     filename?: string | undefined;
+    extracted_text?: string | undefined;
     attachment_id?: string | undefined;
     data_base64?: string | undefined;
     mime_type?: string | undefined;
@@ -3784,6 +3803,6 @@ declare function getPrefetchedSha256(url: string): string | undefined;
  */
 declare function clearPrefetchedShard(url?: string): void;
 
-declare const VERSION = "2026.7.1";
+declare const VERSION = "2026.7.2";
 
 export { type ArchiveInfo, ArchiveManager, type AttachInput, type AttachmentBlobRow, type AttachmentRow, AttachmentsRepository, type BackendCandidate, type BackendCapabilities, type BackendConcept, type BackendLink, type BackendListOptions, type BackendNote, type BackendNoteFull, type BackendProvenanceEdge, type BackendRequest, type BackendSearchHit, type BackendSearchQueryOptions, type BackendSearchResult, type BackendSelection, type BackendSemanticTier, type BackendStartupCost, type BlobStore, type BridgeCapability, type BridgeProviderInfo, type BrowserNoteExport, CURRENT_MIGRATION_HEAD, CURRENT_SHARD_VERSION, type CapabilityInfo, CapabilityManager, type CapabilityName, type CapabilityState, type CaptureKnowledgeInput, CaptureKnowledgeInputSchema, type CaptureKnowledgeResult, type CollectionCreateInput, type CollectionRow, CollectionsRepository, CommunitiesRepository, type CommunityAssignmentView, type CommunityCreateInput, type CommunityFilterDefinition, type CommunityGraph, type CommunityOptions, type CommunitySourceDescriptor, type CommunitySourceType, type CommunitySummary, type CompletionRequest, type CompletionResponse, type ConditionResult, type ConflictStrategy, type CooldownConfig, type CooldownEvent, type CosineSemanticProviderOptions, type CreatePGliteOptions, type CspDirectiveName, type CspDirectives, type CspViolationReport, DB_SNAPSHOT_SCHEMA_VERSION, type DataBackend, type DatabaseClient, type DbSnapshot, type DbSnapshotCompat, type DbSnapshotCompression, type DbSnapshotExpectations, type DbSnapshotMeta, type DbSnapshotSource, DbSnapshotVersionError, type DiscoveredProvider, type DiscoveryOptions, type DumpDbSnapshotOptions, type DumpableDb, EMBED_REQUEST_KIND, EMBED_RESPONSE_KIND, type EmbedFunction, type EmbedRequest, type EmbedRequestMessage, type EmbedResponse, type EmbedResponseMessage, type EmbedTransportPort, type EmbedWorkerOptions, type EmbeddingCompatibilityPolicy, type EmbeddingSetCreateInput, type EmbeddingSetCriteria, type EmbeddingSetDescriptor, type EmbeddingSetEmbeddingInput, type EmbeddingSetFreshness, type EmbeddingSetKind, type EmbeddingSetMode, type EmbeddingSetRow, type EmbeddingSetSelector, EmbeddingSetsRepository, type EnqueueJobInput, type ErrorCategory, type EventMap, type ExportOptions, type FallbackEvent, FallbackRouter, type FallbackRouterConfig, type FortemiBridge, type FortemiBridgeCapabilities, type FortemiBridgeHost, type FortemiConfig, type FortemiCore, type FortemiInferenceRouter, type FortemiSecretStore, type FortemiToolDefinition, FortemiToolManifest, type GetNoteInput, GetNoteInputSchema, type GpuCapabilities, type GraphCommunity, type GraphEdge, type GraphNode, GraphRepository, type IDisposable, type ImportCounts, type ImportOptions, type ImportProgress, type ImportProgressPhase, type ImportResult, type InferenceCapabilities, type InferenceProvider, JOB_CAPABILITIES, JOB_PRIORITIES, type JobQueueOptions, JobQueueWorker, type JobStatus, type JobType, LOCAL_ENDPOINTS, type LinkRow, LinksRepository, type ListNotesInput, ListNotesInputSchema, type LlmCapabilityOptions, type LlmCompleteFn, type LoadedPluginScript, type LocalEndpoint, type ManageArchiveInput, ManageArchiveInputSchema, type ManageArchiveResult, type ManageAttachmentsInput, ManageAttachmentsInputSchema, type ManageAttachmentsResult, type ManageCapabilitiesInput, ManageCapabilitiesInputSchema, type ManageCapabilitiesResult, type ManageCollectionsInput, ManageCollectionsInputSchema, type ManageCollectionsResult, type ManageLinksInput, ManageLinksInputSchema, type ManageLinksResult, type ManageNoteInput, ManageNoteInputSchema, type ManageNoteResult, type ManageTagsInput, ManageTagsInputSchema, type ManageTagsResult, MemoryBlobStore, type Migration, MigrationRunner, type ModelCategory, type ModelFitResult, type ModelInfo, type NoteCreateInput, type NoteFull, type NoteListOptions, type NoteRevision, type NoteSkosTag, type NoteSummary, type NoteUpdateInput, NotesRepository, OpenAICompatibleProvider, type OpenAIProviderConfig, type OpenShardOptions, type PGliteBackendOptions, PGliteStorageBackend, PGliteStorageBackendFactory, PGliteWorkerClient, PGliteWorkerStorageBackend, PGliteWorkerStorageBackendFactory, type PGliteWorkerStorageBackendFactoryOptions, type PaginatedResult, type PersistenceMode, type PluginCspOptions, type PluginScriptDescriptor, type PluginScriptPolicy, type PrefetchOptions, type PrefetchResult, type ProbeResult, type ProbeStatus, type ProvenanceEdge, ProvenanceRepository, type ProviderCapabilities, ProviderRegistry, type ProviderTier, type QueryExecutor, type QueryResult, type RecommendedTier, type RecordProvenanceInput, type RemoteBackendConfig, type RemoteBackendPaths, type ResolvedEmbeddingRow, type ResolvedEmbeddingSet, type RestoreDbSnapshotOptions, type RouteHandler, SHARD_FORMAT, SUPPORTED_PGLITE_VERSION, type SWRegistrationResult, type SearchFacets, type SearchInput, SearchInputSchema, type SearchOptions, SearchRepository, type SearchResponse, type SearchResult, type ShardBackendOptions, type ShardClusterRef, type ShardCollection, type ShardComponent, type ShardComponentStore, type ShardEmbedding, type ShardEmbeddingSet, type ShardEmbeddingSetMember, type ShardLayout, type ShardLink, type ShardListOptions, type ShardManifest, type ShardNote, type ShardNoteFull, type ShardNoteSkosTag, type ShardProvenanceEdge, type ShardReader, type ShardReaderNote, type ShardReaderSource, type ShardSearchOptions, type ShardSearchRankedNote, type ShardSearchResult, type ShardSearchWeights, type ShardSkosConcept, type ShardSkosRelation, type ShardSkosScheme, type ShardTag, type SimilarityGraphCacheKey, type SimilarityGraphOptions, type SimilarityGraphRequest, type SimilarityGraphResult, type SkosConcept, type SkosRelation, SkosRepository, type SkosScheme, type StaticSemanticProvider, type StorageBackend, type StorageBackendFactory, type StorageOpenRequest, type StorageTopology, type StreamChunk, TagsRepository, TransactionProxy, TypedEventBus, VERSION, type VectorEntry, type VirtualEmbeddingSetDefinition, type VirtualEmbeddingSetSource, type VirtualEmbeddingSetValidationError, type VirtualMaterializationPolicy, type VramTier, type WorkerRequest, type WorkerResponse, aiRevisionHandler, allMigrations, appendPluginScript, buildNoteConditions, buildPluginCsp, captureKnowledge, chunkText, classifyError, classifyModel, clearPrefetchedShard, collectionFromShard, collectionToShard, computeHash, computeSri, conceptTaggingHandler, cosineSimilarity, createBlobStore, createCosineSemanticProvider, createCspReportHandler, createFortemi, createLegacyProvider, createPGliteBackend, createPGliteInstance, createRemoteBackend, createRoutes, createShardBackend, createWorkerEmbedFunction, defaultStorageBackendFactory, detectCommunities, detectGpuCapabilities, detectInferenceCapabilities, discoverLocalProviders, dumpDbSnapshot, embeddingFromShard, embeddingGenerationHandler, embeddingSetFromShard, embeddingSetMemberToShard, embeddingSetToShard, embeddingToShard, enqueueFullWorkflow, enqueueJob, enqueueNoteCreationJobs, estimateModelFit, estimateVramMB, estimateVramTier, exportShard, fetchPluginScript, fortemiManifest, fromPrefetched, generateId, getEmbedFunction, getFortemiBridge, getFortemiSecretStore, getJobQueueStatus, getLlmFunction, getNote, getPrefetchedSha256, handleEmbedRequests, hasFortemiSecureSecrets, importShard, isPluginScriptAllowed, isShardPrefetched, linkFromShard, linkToShard, linkingHandler, listNotes, manageArchive, manageAttachments, manageCapabilities, manageCollections, manageLinks, manageNote, manageTags, matchRoute, noteFromShard, noteSkosTagToShard, noteToShard, openShard, packTarGz, parseCspReport, prefetchShard, provenanceEdgeToShard, registerLlmCapability, registerSemanticCapability, registerSemanticCapabilityWorker, registerServiceWorker, restoreDbSnapshot, searchTool, selectBackend, selectLlmModel, setEmbedFunction, setLlmFunction, sha256Hex, skosConceptToShard, skosRelationToShard, skosSchemeToShard, suggestTags, tagsFromShard, tagsToShard, titleGenerationHandler, unpackTarGz, unregisterLlmCapability, unregisterSemanticCapability, validateChecksums, verifyDbSnapshotMeta, verifySri };

package/dist/index.js

@@ -596,6 +596,19 @@ var migration0009 = {
   `
 };
 
+// src/migrations/0010_attachment_text_metadata.ts
+var migration0010 = {
+  version: 10,
+  name: "0010_attachment_text_metadata",
+  sql: `
+    ALTER TABLE attachment
+      ADD COLUMN IF NOT EXISTS mime_type TEXT;
+
+    ALTER TABLE attachment
+      ADD COLUMN IF NOT EXISTS extracted_text TEXT;
+  `
+};
+
 // src/migrations/index.ts
 var allMigrations = [
   migration0001,
@@ -606,7 +619,8 @@ var allMigrations = [
   migration0006,
   migration0007,
   migration0008,
-  migration0009
+  migration0009,
+  migration0010
 ];
 
 // src/data-archive.ts
@@ -1119,6 +1133,16 @@ function buildNoteConditions(options, startIdx, includeDeleted = false) {
 }
 
 // src/repositories/embedding-sets-repository.ts
+var ATTACHMENT_TEXT_JOIN = `
+       LEFT JOIN (
+         SELECT note_id,
+                string_agg(extracted_text, ' ' ORDER BY position, created_at)
+                  FILTER (WHERE extracted_text IS NOT NULL AND extracted_text <> '') as extracted_text
+         FROM attachment
+         WHERE deleted_at IS NULL
+         GROUP BY note_id
+       ) ax ON ax.note_id = n.id`;
+var COMBINED_TEXT_VECTOR_SQL = `to_tsvector('english', (coalesce(c.content, '') || ' ' || coalesce(ax.extracted_text, '')))`;
 var DEFAULT_COMPATIBILITY = {
   model: "require-same",
   dimension: "require-same",
@@ -1498,7 +1522,7 @@ var EmbeddingSetsRepository = class {
       params.push(criteria.updatedBefore);
     }
     if (criteria.query?.trim()) {
-      conditions.push(`(n.tsv @@ plainto_tsquery('english', $${idx}) OR to_tsvector('english', coalesce(c.content, '')) @@ plainto_tsquery('english', $${idx}))`);
+      conditions.push(`(n.tsv @@ plainto_tsquery('english', $${idx}) OR ${COMBINED_TEXT_VECTOR_SQL} @@ plainto_tsquery('english', $${idx}))`);
       params.push(criteria.query);
     }
     const result = await this.db.query(
@@ -1506,6 +1530,7 @@ var EmbeddingSetsRepository = class {
        FROM embedding e
        JOIN note n ON n.id = e.note_id
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN}
        WHERE ${conditions.join(" AND ")}
        ORDER BY e.note_id, e.created_at DESC`,
       params
@@ -1623,6 +1648,17 @@ var EmbeddingSetsRepository = class {
 };
 
 // src/repositories/search-repository.ts
+var ATTACHMENT_TEXT_JOIN2 = `
+       LEFT JOIN (
+         SELECT note_id,
+                string_agg(extracted_text, ' ' ORDER BY position, created_at)
+                  FILTER (WHERE extracted_text IS NOT NULL AND extracted_text <> '') as extracted_text
+         FROM attachment
+         WHERE deleted_at IS NULL
+         GROUP BY note_id
+       ) ax ON ax.note_id = n.id`;
+var COMBINED_TEXT_SQL = `(coalesce(c.content, '') || ' ' || coalesce(ax.extracted_text, ''))`;
+var COMBINED_TEXT_VECTOR_SQL2 = `to_tsvector('english', ${COMBINED_TEXT_SQL})`;
 var SearchRepository = class {
   constructor(db, semanticAvailable = false) {
     this.db = db;
@@ -1707,7 +1743,7 @@ var SearchRepository = class {
     const { conditions, params, nextIdx } = buildNoteConditions(options, 2);
     conditions.unshift(
       `(n.tsv @@ ${tsqFn}('english', $1) OR
-        to_tsvector('english', coalesce(c.content, '')) @@ ${tsqFn}('english', $1))`
+        ${COMBINED_TEXT_VECTOR_SQL2} @@ ${tsqFn}('english', $1))`
     );
     let paramIdx = this.scopeToResolvedEmbeddingSet(conditions, params, nextIdx, resolvedEmbeddingSet);
     const allParams = [query, ...params];
@@ -1716,6 +1752,7 @@ var SearchRepository = class {
       `SELECT COUNT(*) as count
        FROM note n
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE ${where}`,
       allParams
     );
@@ -1724,17 +1761,18 @@ var SearchRepository = class {
     const result = await this.db.query(
       `SELECT n.id, n.title, n.created_at, n.updated_at,
               ts_rank(
-                setweight(n.tsv, 'A') || setweight(to_tsvector('english', coalesce(c.content, '')), 'B'),
+                setweight(n.tsv, 'A') || setweight(${COMBINED_TEXT_VECTOR_SQL2}, 'B'),
                 ${tsqFn}('english', $1)
               ) as rank,
               ts_headline(
                 'english',
-                coalesce(c.content, ''),
+                ${COMBINED_TEXT_SQL},
                 ${tsqFn}('english', $1),
                 'StartSel=<mark>, StopSel=</mark>, MaxWords=35, MinWords=15'
               ) as snippet
        FROM note n
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE ${where}
        ORDER BY rank DESC, n.created_at DESC
        LIMIT $${paramIdx++} OFFSET $${paramIdx++}`,
@@ -1750,6 +1788,7 @@ var SearchRepository = class {
       const idsResult = await this.db.query(
         `SELECT n.id FROM note n
          LEFT JOIN note_revised_current c ON c.note_id = n.id
+         ${ATTACHMENT_TEXT_JOIN2}
          WHERE ${where}`,
         allParams
       );
@@ -1796,10 +1835,11 @@ var SearchRepository = class {
     const result = await this.db.query(
       `SELECT n.id, n.title, n.created_at, n.updated_at,
               (e.vector <=> $${vecIdx}::vector) as distance,
-              LEFT(coalesce(c.content, ''), 200) as snippet
+              LEFT(${COMBINED_TEXT_SQL}, 200) as snippet
        FROM embedding e
        JOIN note n ON n.id = e.note_id
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE ${where}
        ORDER BY e.vector <=> $${vecIdx}::vector ASC
        LIMIT $${limIdx} OFFSET $${offIdx}`,
@@ -1840,7 +1880,7 @@ var SearchRepository = class {
     const textConditions = [
       ...textCond.conditions,
       `(n.tsv @@ ${tsqFn}('english', $1) OR
-        to_tsvector('english', coalesce(c.content, '')) @@ ${tsqFn}('english', $1))`
+        ${COMBINED_TEXT_VECTOR_SQL2} @@ ${tsqFn}('english', $1))`
     ];
     this.scopeToResolvedEmbeddingSet(textConditions, textCond.params, textCond.nextIdx, resolvedEmbeddingSet);
     const textWhere = textConditions.join(" AND ");
@@ -1848,11 +1888,12 @@ var SearchRepository = class {
     const textResult = await this.db.query(
       `SELECT n.id,
               ts_rank(
-                setweight(n.tsv, 'A') || setweight(to_tsvector('english', coalesce(c.content, '')), 'B'),
+                setweight(n.tsv, 'A') || setweight(${COMBINED_TEXT_VECTOR_SQL2}, 'B'),
                 ${tsqFn}('english', $1)
               ) as rank
        FROM note n
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE ${textWhere}
        ORDER BY rank DESC
        LIMIT 100`,
@@ -1886,9 +1927,10 @@ var SearchRepository = class {
     }
     const noteResult = await this.db.query(
       `SELECT n.id, n.title, n.created_at, n.updated_at,
-              LEFT(coalesce(c.content, ''), 200) as snippet
+              LEFT(${COMBINED_TEXT_SQL}, 200) as snippet
        FROM note n
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE n.id = ANY($1)`,
       [pageIds]
     );
@@ -1936,9 +1978,10 @@ var SearchRepository = class {
     const listParams = [...params, limit, offset];
     const result = await this.db.query(
       `SELECT n.id, n.title, n.created_at, n.updated_at,
-              LEFT(coalesce(c.content, ''), 200) as snippet
+              LEFT(${COMBINED_TEXT_SQL}, 200) as snippet
        FROM note n
        LEFT JOIN note_revised_current c ON c.note_id = n.id
+       ${ATTACHMENT_TEXT_JOIN2}
        WHERE ${where}
        ORDER BY n.created_at DESC
        LIMIT $${paramIdx++} OFFSET $${paramIdx++}`,
@@ -3811,6 +3854,29 @@ function getLlmFunction() {
   return llmFn;
 }
 
+// src/repositories/note-text.ts
+async function getNoteTextWithExtractedAttachments(db, noteId) {
+  const result = await db.query(
+    `SELECT c.content,
+            COALESCE(string_agg(a.extracted_text, E'
+' ORDER BY a.position, a.created_at)
+              FILTER (WHERE a.extracted_text IS NOT NULL AND a.extracted_text <> ''), '') as extracted_text
+       FROM note_revised_current c
+       LEFT JOIN attachment a ON a.note_id = c.note_id AND a.deleted_at IS NULL
+       WHERE c.note_id = $1
+       GROUP BY c.content`,
+    [noteId]
+  );
+  if (result.rows.length === 0) return null;
+  const content = result.rows[0].content;
+  const extractedText = result.rows[0].extracted_text ?? "";
+  return {
+    content,
+    extracted_text: extractedText,
+    combined: [content, extractedText].filter(Boolean).join("\n")
+  };
+}
+
 // src/job-queue-worker.ts
 var JOB_PRIORITIES = {
   ai_revision: 1,
@@ -4092,12 +4158,9 @@ function conceptTaggingHandler(job, db) {
   return (async () => {
     const llmFn2 = getLlmFunction();
     if (!llmFn2) return { skipped: true, reason: "no LLM function registered" };
-    const result = await db.query(
-      `SELECT content FROM note_revised_current WHERE note_id = $1`,
-      [job.note_id]
-    );
-    if (result.rows.length === 0) throw new Error(`No content found for note ${job.note_id}`);
-    const content = result.rows[0].content;
+    const noteText = await getNoteTextWithExtractedAttachments(db, job.note_id);
+    if (!noteText) throw new Error(`No content found for note ${job.note_id}`);
+    const content = noteText.combined;
     const prompt = `Task: Extract 3-5 topic tags from the text below.
 Rules:
 - Each tag is 1-3 words, lowercase
@@ -4990,9 +5053,17 @@ var AttachmentsRepository = class {
     }
     const attachmentId = generateId();
     await this.db.query(
-      `INSERT INTO attachment (id, note_id, blob_id, filename, display_name)
-       VALUES ($1, $2, $3, $4, $5)`,
-      [attachmentId, input.noteId, blobId, input.filename, input.displayName ?? null]
+      `INSERT INTO attachment (id, note_id, blob_id, filename, display_name, mime_type, extracted_text)
+       VALUES ($1, $2, $3, $4, $5, $6, $7)`,
+      [
+        attachmentId,
+        input.noteId,
+        blobId,
+        input.filename,
+        input.displayName ?? null,
+        input.mimeType ?? null,
+        input.extractedText ?? null
+      ]
     );
     return this.get(attachmentId);
   }
@@ -5055,6 +5126,7 @@ var ManageAttachmentsInputSchema = z.object({
   data_base64: z.string().optional(),
   filename: z.string().optional(),
   mime_type: z.string().optional(),
+  extracted_text: z.string().optional(),
   display_name: z.string().optional()
 });
 async function manageAttachments(db, blobStore, rawInput) {
@@ -5073,6 +5145,7 @@ async function manageAttachments(db, blobStore, rawInput) {
         data,
         filename: input.filename,
         mimeType: input.mime_type,
+        extractedText: input.extracted_text,
         displayName: input.display_name
       });
       return { action: "attach", attachment, size_bytes: data.length };
@@ -5324,12 +5397,9 @@ function averageEmbeddings(embeddings) {
 async function embeddingGenerationHandler(job, db) {
   const fn = embedFn;
   if (!fn) return { skipped: true, reason: "no embed function registered" };
-  const noteResult = await db.query(
-    `SELECT content FROM note_revised_current WHERE note_id = $1`,
-    [job.note_id]
-  );
-  if (noteResult.rows.length === 0) throw new Error(`No content for note ${job.note_id}`);
-  const content = noteResult.rows[0].content;
+  const noteText = await getNoteTextWithExtractedAttachments(db, job.note_id);
+  if (!noteText) throw new Error(`No content for note ${job.note_id}`);
+  const content = noteText.combined;
   const chunks = chunkText(content);
   const embeddings = await fn(chunks);
   const vector2 = averageEmbeddings(embeddings);
@@ -6512,6 +6582,7 @@ function noteToShard(note) {
     title: note.title,
     original_content: note.original_content,
     revised_content: note.revised_content,
+    ...note.binary_sources?.length ? { binary_sources: note.binary_sources } : {},
     format: note.format,
     source: note.source,
     starred: note.is_starred,
@@ -6532,6 +6603,7 @@ function noteFromShard(shard) {
     is_archived: shard.archived,
     original_content: shard.original_content,
     revised_content: shard.revised_content,
+    binary_sources: shard.binary_sources,
     tags: shard.tags,
     created_at: shard.created_at,
     updated_at: shard.updated_at,
@@ -6789,9 +6861,40 @@ async function exportShard(db, options) {
     tags.push(row.tag);
     tagsByNote.set(row.note_id, tags);
   }
+  const attachmentRows = await db.query(
+    `SELECT a.note_id,
+            a.id,
+            a.filename,
+            a.mime_type,
+            a.extracted_text,
+            b.content_hash,
+            b.size_bytes,
+            b.storage_path
+       FROM attachment a
+       JOIN attachment_blob b ON b.id = a.blob_id
+       WHERE a.deleted_at IS NULL
+       ORDER BY a.note_id, a.position, a.created_at`
+  );
+  const binarySourcesByNote = /* @__PURE__ */ new Map();
+  for (const row of attachmentRows.rows) {
+    const source = {
+      extracted_text: row.extracted_text ?? "",
+      attachment: {
+        id: row.id,
+        path: row.storage_path ?? row.filename,
+        mime: row.mime_type,
+        checksum: row.content_hash,
+        bytes: Number(row.size_bytes)
+      }
+    };
+    const sources = binarySourcesByNote.get(row.note_id) ?? [];
+    sources.push(source);
+    binarySourcesByNote.set(row.note_id, sources);
+  }
   const notes = noteRows.rows.map((row) => ({
     ...row,
-    tags: tagsByNote.get(row.id) ?? []
+    tags: tagsByNote.get(row.id) ?? [],
+    binary_sources: binarySourcesByNote.get(row.id)
   }));
   const exportedNoteIds = new Set(notes.map((n) => n.id));
   const shardNotes = notes.map((n) => noteToShard(n));
@@ -7653,7 +7756,8 @@ function tokenize(query) {
   return query.toLowerCase().split(/[^a-z0-9]+/i).filter((token) => token.length > 0);
 }
 function noteSearchText(note) {
-  return `${note.title ?? ""} ${note.original_content} ${note.revised_content ?? ""}`.toLowerCase();
+  const extractedText = note.binary_sources?.map((source) => source.extracted_text).filter(Boolean).join(" ") ?? "";
+  return `${note.title ?? ""} ${note.original_content} ${note.revised_content ?? ""} ${extractedText}`.toLowerCase();
 }
 function countOccurrences(haystack, needle) {
   if (!needle) return 0;
@@ -7674,7 +7778,8 @@ function noteMatchesTokens(note, tokens) {
 function rankNote(note, tokens, weights) {
   if (tokens.length === 0) return 0;
   const title = (note.title ?? "").toLowerCase();
-  const content = `${note.original_content} ${note.revised_content ?? ""}`.toLowerCase();
+  const extractedText = note.binary_sources?.map((source) => source.extracted_text).filter(Boolean).join(" ") ?? "";
+  const content = `${note.original_content} ${note.revised_content ?? ""} ${extractedText}`.toLowerCase();
   const tagText = note.tags.join(" ").toLowerCase();
   let score = 0;
   for (const token of tokens) {
@@ -7685,7 +7790,8 @@ function rankNote(note, tokens, weights) {
   return score;
 }
 function makeSnippet(note, tokens, length) {
-  const content = `${note.original_content} ${note.revised_content ?? ""}`.trim();
+  const extractedText = note.binary_sources?.map((source) => source.extracted_text).filter(Boolean).join(" ") ?? "";
+  const content = `${note.original_content} ${note.revised_content ?? ""} ${extractedText}`.trim();
   if (tokens.length === 0) return content.slice(0, length);
   const lower = content.toLowerCase();
   const firstAt = tokens.map((token) => lower.indexOf(token)).filter((index) => index !== -1).sort((a, b) => a - b)[0] ?? 0;
@@ -8470,7 +8576,55 @@ function recordTitle(item) {
   return item.title ?? item.search?.title ?? item.search?.name ?? item.id;
 }
 function recordText(item) {
-  return item.text ?? item.search?.body ?? item.search?.summary ?? item.chunks?.map((chunk) => chunk.text ?? chunk.body ?? chunk.summary ?? "").filter(Boolean).join("\n") ?? "";
+  const base = item.text ?? item.search?.body ?? item.search?.summary ?? item.chunks?.map((chunk) => chunk.text ?? chunk.body ?? chunk.summary ?? "").filter(Boolean).join("\n") ?? "";
+  const extractedText = "binary_sources" in item ? item.binary_sources?.map((source) => source.extracted_text).filter(Boolean).join("\n") ?? "" : "";
+  return [base, extractedText].filter(Boolean).join("\n");
+}
+function defaultEmbeddingInput(record, granularity) {
+  const title = recordTitle(record);
+  const text = recordText(record);
+  if (granularity === "title-summary") {
+    const extractedText = record.binary_sources?.map((source) => source.extracted_text).filter(Boolean).join("\n") ?? "";
+    return [title, record.search?.summary ?? "", extractedText].filter(Boolean).join("\n");
+  }
+  return [title, text].filter(Boolean).join("\n");
+}
+function generatedAtString(value) {
+  if (value instanceof Date) return value.toISOString();
+  return value ?? (/* @__PURE__ */ new Date()).toISOString();
+}
+async function buildAiwgStaticEmbeddingSet(index, options) {
+  assertAiwgFortemiIndexExport(index);
+  const granularity = options.granularity ?? "body";
+  const records = options.records ?? index.items;
+  const embeddings = [];
+  for (const record of records) {
+    const input = options.textForRecord?.(record) ?? defaultEmbeddingInput(record, granularity);
+    const embedding = await options.backend.embed(input, record);
+    if (embedding.length !== options.backend.dimensions) {
+      throw new Error(`Embedding for ${record.id} has ${embedding.length} dimensions; expected ${options.backend.dimensions}`);
+    }
+    embeddings.push({
+      record_id: record.id,
+      embedding,
+      granularity,
+      input_hash: computeHash(new TextEncoder().encode(input)),
+      source_path: record.source.path
+    });
+  }
+  const embeddingSet = {
+    schema_version: "aiwg.fortemi.embedding.set.v1",
+    id: options.id,
+    model: options.backend.model,
+    dimensions: options.backend.dimensions,
+    generated_at: generatedAtString(options.generatedAt),
+    granularity,
+    ...options.metric ? { metric: options.metric } : {},
+    input_hash_algorithm: "sha256",
+    embeddings
+  };
+  assertAiwgStaticEmbeddingSet(embeddingSet);
+  return embeddingSet;
 }
 function recordSearchValues(item) {
   const search = item.search;
@@ -9364,8 +9518,8 @@ function communityIdsFor(item, options) {
 }
 
 // src/index.ts
-var VERSION = "2026.7.1";
+var VERSION = "2026.7.2";
 
-export { AIWG_SCAN_REQUIRED_FIELDS, ArchiveManager, AttachmentsRepository, CURRENT_MIGRATION_HEAD, CURRENT_SHARD_VERSION, CapabilityManager, CaptureKnowledgeInputSchema, CollectionsRepository, CommunitiesRepository, DB_SNAPSHOT_SCHEMA_VERSION, DbSnapshotVersionError, EMBED_REQUEST_KIND, EMBED_RESPONSE_KIND, EmbeddingSetsRepository, FallbackRouter, FortemiToolManifest, GetNoteInputSchema, GraphRepository, JOB_CAPABILITIES, JOB_PRIORITIES, JobQueueWorker, LOCAL_ENDPOINTS, LinksRepository, ListNotesInputSchema, ManageArchiveInputSchema, ManageAttachmentsInputSchema, ManageCapabilitiesInputSchema, ManageCollectionsInputSchema, ManageLinksInputSchema, ManageNoteInputSchema, ManageTagsInputSchema, MemoryBlobStore, MigrationRunner, NotesRepository, OpenAICompatibleProvider, PGliteStorageBackend, PGliteStorageBackendFactory, PGliteWorkerClient, PGliteWorkerStorageBackend, PGliteWorkerStorageBackendFactory, ProvenanceRepository, ProviderRegistry, SHARD_FORMAT, SUPPORTED_PGLITE_VERSION, SearchInputSchema, SearchRepository, SkosRepository, TagsRepository, TransactionProxy, TypedEventBus, VERSION, aiRevisionHandler, aiwgFortemiIndexToCommunityGraph, allMigrations, appendPluginScript, assertAiwgFortemiChunkManifest, assertAiwgFortemiChunkPart, assertAiwgFortemiIndexExport, assertAiwgStaticEmbeddingSet, buildAiwgChunkedIndex, buildNoteConditions, buildPluginCsp, captureKnowledge, chunkText, classifyError, classifyModel, clearPrefetchedShard, collectionFromShard, collectionToShard, computeHash, computeSri, conceptTaggingHandler, cosineSimilarity, createAiwgFetchChunkLoader, createAiwgFetchDetailLoader, createAiwgIndexController, createAiwgReviewDecisionExport, createBlobStore, createCosineSemanticProvider, createCspReportHandler, createFortemi, createLegacyProvider, createPGliteBackend, createPGliteInstance, createRemoteBackend, createRoutes, createShardBackend, createWorkerEmbedFunction, defaultStorageBackendFactory, detectCommunities, detectGpuCapabilities, detectInferenceCapabilities, discoverLocalProviders, dumpDbSnapshot, embeddingFromShard, embeddingGenerationHandler, embeddingSetFromShard, embeddingSetMemberToShard, embeddingSetToShard, embeddingToShard, enqueueFullWorkflow, enqueueJob, enqueueNoteCreationJobs, estimateModelFit, estimateVramMB, estimateVramTier, exportShard, fetchPluginScript, findAiwgStaticDuplicatePairs, fortemiManifest, fromPrefetched, generateId, getAiwgFortemiFacets, getEmbedFunction, getFortemiBridge, getFortemiSecretStore, getJobQueueStatus, getLlmFunction, getNote, getPrefetchedSha256, handleEmbedRequests, hasFortemiSecureSecrets, importShard, isPluginScriptAllowed, isShardPrefetched, linkFromShard, linkToShard, linkingHandler, listNotes, manageArchive, manageAttachments, manageCapabilities, manageCollections, manageLinks, manageNote, manageTags, matchRoute, noteFromShard, noteSkosTagToShard, noteToShard, openShard, packTarGz, parseCspReport, prefetchShard, provenanceEdgeToShard, queryAiwgFortemiIndex, queryAiwgHybridIndex, queryAiwgSemanticIndex, registerLlmCapability, registerSemanticCapability, registerSemanticCapabilityWorker, registerServiceWorker, restoreDbSnapshot, searchTool, selectBackend, selectLlmModel, setEmbedFunction, setLlmFunction, sha256Hex, skosConceptToShard, skosRelationToShard, skosSchemeToShard, suggestTags, tagsFromShard, tagsToShard, titleGenerationHandler, unpackTarGz, unregisterLlmCapability, unregisterSemanticCapability, validateAiwgFortemiChunkManifest, validateAiwgFortemiChunkPart, validateAiwgFortemiIndexExport, validateAiwgStaticEmbeddingSet, validateChecksums, verifyDbSnapshotMeta, verifySri };
+export { AIWG_SCAN_REQUIRED_FIELDS, ArchiveManager, AttachmentsRepository, CURRENT_MIGRATION_HEAD, CURRENT_SHARD_VERSION, CapabilityManager, CaptureKnowledgeInputSchema, CollectionsRepository, CommunitiesRepository, DB_SNAPSHOT_SCHEMA_VERSION, DbSnapshotVersionError, EMBED_REQUEST_KIND, EMBED_RESPONSE_KIND, EmbeddingSetsRepository, FallbackRouter, FortemiToolManifest, GetNoteInputSchema, GraphRepository, JOB_CAPABILITIES, JOB_PRIORITIES, JobQueueWorker, LOCAL_ENDPOINTS, LinksRepository, ListNotesInputSchema, ManageArchiveInputSchema, ManageAttachmentsInputSchema, ManageCapabilitiesInputSchema, ManageCollectionsInputSchema, ManageLinksInputSchema, ManageNoteInputSchema, ManageTagsInputSchema, MemoryBlobStore, MigrationRunner, NotesRepository, OpenAICompatibleProvider, PGliteStorageBackend, PGliteStorageBackendFactory, PGliteWorkerClient, PGliteWorkerStorageBackend, PGliteWorkerStorageBackendFactory, ProvenanceRepository, ProviderRegistry, SHARD_FORMAT, SUPPORTED_PGLITE_VERSION, SearchInputSchema, SearchRepository, SkosRepository, TagsRepository, TransactionProxy, TypedEventBus, VERSION, aiRevisionHandler, aiwgFortemiIndexToCommunityGraph, allMigrations, appendPluginScript, assertAiwgFortemiChunkManifest, assertAiwgFortemiChunkPart, assertAiwgFortemiIndexExport, assertAiwgStaticEmbeddingSet, buildAiwgChunkedIndex, buildAiwgStaticEmbeddingSet, buildNoteConditions, buildPluginCsp, captureKnowledge, chunkText, classifyError, classifyModel, clearPrefetchedShard, collectionFromShard, collectionToShard, computeHash, computeSri, conceptTaggingHandler, cosineSimilarity, createAiwgFetchChunkLoader, createAiwgFetchDetailLoader, createAiwgIndexController, createAiwgReviewDecisionExport, createBlobStore, createCosineSemanticProvider, createCspReportHandler, createFortemi, createLegacyProvider, createPGliteBackend, createPGliteInstance, createRemoteBackend, createRoutes, createShardBackend, createWorkerEmbedFunction, defaultStorageBackendFactory, detectCommunities, detectGpuCapabilities, detectInferenceCapabilities, discoverLocalProviders, dumpDbSnapshot, embeddingFromShard, embeddingGenerationHandler, embeddingSetFromShard, embeddingSetMemberToShard, embeddingSetToShard, embeddingToShard, enqueueFullWorkflow, enqueueJob, enqueueNoteCreationJobs, estimateModelFit, estimateVramMB, estimateVramTier, exportShard, fetchPluginScript, findAiwgStaticDuplicatePairs, fortemiManifest, fromPrefetched, generateId, getAiwgFortemiFacets, getEmbedFunction, getFortemiBridge, getFortemiSecretStore, getJobQueueStatus, getLlmFunction, getNote, getPrefetchedSha256, handleEmbedRequests, hasFortemiSecureSecrets, importShard, isPluginScriptAllowed, isShardPrefetched, linkFromShard, linkToShard, linkingHandler, listNotes, manageArchive, manageAttachments, manageCapabilities, manageCollections, manageLinks, manageNote, manageTags, matchRoute, noteFromShard, noteSkosTagToShard, noteToShard, openShard, packTarGz, parseCspReport, prefetchShard, provenanceEdgeToShard, queryAiwgFortemiIndex, queryAiwgHybridIndex, queryAiwgSemanticIndex, registerLlmCapability, registerSemanticCapability, registerSemanticCapabilityWorker, registerServiceWorker, restoreDbSnapshot, searchTool, selectBackend, selectLlmModel, setEmbedFunction, setLlmFunction, sha256Hex, skosConceptToShard, skosRelationToShard, skosSchemeToShard, suggestTags, tagsFromShard, tagsToShard, titleGenerationHandler, unpackTarGz, unregisterLlmCapability, unregisterSemanticCapability, validateAiwgFortemiChunkManifest, validateAiwgFortemiChunkPart, validateAiwgFortemiIndexExport, validateAiwgStaticEmbeddingSet, validateChecksums, verifyDbSnapshotMeta, verifySri };
 //# sourceMappingURL=index.js.map
 //# sourceMappingURL=index.js.map