@prmichaelsen/remember-mcp 2.7.7 → 2.7.9

package/CHANGELOG.md

@@ -5,6 +5,60 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## [2.7.8] - 2026-02-17
+
+### Fixed
+
+- **CRITICAL: Fixed Shared Space Discovery Bug**
+  - Search and query tools now correctly filter by `doc_type: 'memory'` instead of `'space_memory'`
+  - The `space_memory` concept was removed from the architecture
+  - Search tools were filtering for `doc_type: 'space_memory'` but published memories had `doc_type: 'memory'`
+  - Now all users can discover memories published to shared spaces like The Void
+  - Root cause: Lines 159 in `search-space.ts` and 147 in `query-space.ts` filtered for wrong doc_type
+  - Fix: Changed filters from `'space_memory'` to `'memory'` to match actual data
+
+### Technical Details
+
+- Modified: `src/tools/search-space.ts` (line 159)
+- Modified: `src/tools/query-space.ts` (line 147)
+- Changed filter from `.equal('space_memory')` to `.equal('memory')`
+- Published memories keep `doc_type: 'memory'` (as intended)
+
+### Impact
+
+- **Fixes**: Shared space discovery now works correctly
+- **Unblocks**: Core shared spaces functionality (The Void)
+- **Note**: No data migration needed - existing published memories are correct
+
+---
+
+## [2.7.7] - 2026-02-17
+
+### Fixed
+
+- **CRITICAL: Fixed Memory_public vectorizer missing**
+  - `ensurePublicCollection()` now properly creates collection with vectorizer
+  - Previously relied on auto-schema which doesn't include vectorizers
+  - Now calls `createSpaceCollection(client, 'public')` to create with full schema
+  - Fixes "vectorizer not configured" errors when searching The Void
+
+### Root Cause
+
+- Commit bcde314 changed to auto-schema approach for schema compatibility
+- Auto-schema creates collections without vectorizers (vectorizer: "none")
+- This broke semantic search in Memory_public (The Void)
+- Manual creation with vectorizer config is required
+
+### Technical Details
+
+- Modified: `src/weaviate/space-schema.ts` (lines 382-398)
+- Changed `ensurePublicCollection()` to check if collection exists
+- If not exists, calls `createSpaceCollection(client, 'public')`
+- Creates collection with vectorizer: `text2VecOpenAI`, model: `text-embedding-3-small`
+- Vectorizes: `content`, `title`, `summary`, `observation`
+
+---
+
 ## [2.7.6] - 2026-02-17
 
 ### Changed

package/agent/progress.yaml

@@ -2,11 +2,11 @@
 
 project:
   name: remember-mcp
-  version: 2.6.3
+  version: 2.7.8
   started: 2026-02-11
   status: in_progress
   current_milestone: M12
-  last_updated: 2026-02-16
+  last_updated: 2026-02-17
 
 milestones:
   - id: M1
@@ -523,6 +523,19 @@ progress:
   overall: 50%
 
 recent_work:
+  - date: 2026-02-17
+    description: CRITICAL BUG FIX - Shared Space Discovery Now Works (v2.7.8)
+    items:
+      - 🐛 CRITICAL: Fixed shared space discovery bug
+      - ✅ Search/query tools now filter by doc_type: 'memory' (was 'space_memory')
+      - ✅ space_memory concept was removed from architecture
+      - ✅ All users can now discover memories in The Void
+      - ✅ Root cause: search-space.ts and query-space.ts filtered for wrong doc_type
+      - ✅ Changed filters from 'space_memory' to 'memory'
+      - ✅ Build successful, all 81 tests passing
+      - ✅ No data migration needed - existing memories are correct
+      - 📋 Released v2.7.8 with critical bug fix
+      
   - date: 2026-02-17
     description: Fixed Build Error & Completed Task 66 - Comment Field Initialization
     items:
@@ -544,20 +557,25 @@ recent_work:
       - 📋 Ready for M12 completion: Tasks 58-59 remaining
       
   - date: 2026-02-17
-    description: ACP Initialization Complete - Project Context Loaded
+    description: ACP Initialization Complete - Ready for M12 Completion
     items:
       - ✅ ACP initialization workflow executed via @acp.init command
       - ⚠️  ACP updates available (v1.4.3 with improved script color output)
-      - ✅ All agent documentation reviewed (23 design docs, 11 milestones, 66 tasks, 5 patterns)
+      - ✅ All agent documentation reviewed (30 design docs, 12 milestones, 67 tasks, 7 patterns)
       - ✅ Project status verified: M1-M4, M10-M11 complete (100%), M12 in progress (60%)
-      - ✅ Version 2.6.7 confirmed (latest release)
-      - ✅ 53 unit tests passing (1 skipped integration test)
-      - ⚠️  TypeScript compilation error: Missing 'poetry' in content-types.ts
+      - ✅ Version 2.7.7 confirmed (latest release)
+      - ✅ 81 unit tests passing (1 skipped integration test)
+      - ✅ Test coverage: 33.12% overall
+      - ✅ TypeScript compiles without errors
+      - ✅ Build successful (dual export: server + factory)
       - ✅ All 17 MCP tools implemented and working
       - ✅ M12 Comment System: 3/5 tasks complete (schema, search, query updates)
+      - ✅ Task 66 completed: Comment fields initialized in create/update tools
       - 📋 Remaining M12 tasks: Task 58 (unit tests), Task 59 (documentation)
-      - 📋 Build blocked by TypeScript error in content-types.ts
-      - 📋 Need to fix 'poetry' content type metadata
+      - 📋 Ready to complete M12 milestone
+      - 📋 Multi-tenant architecture with per-user isolation operational
+      - 📋 Vector search with Weaviate + metadata with Firestore working
+      - 📋 Shared spaces with multi-space support functional
       
   - date: 2026-02-16
     description: Comment System, Structured Logging, and Safety Guidelines (v2.6.0-v2.6.2)

package/dist/server-factory.js

@@ -3688,11 +3688,284 @@ init_space_memory();
 init_logger();
 import weaviate3 from "weaviate-client";
 var PUBLIC_COLLECTION_NAME = "Memory_public";
+function getSpaceCollectionName(spaceId) {
+  return `Memory_${spaceId}`;
+}
 function isValidSpaceId(spaceId) {
   return SUPPORTED_SPACES.includes(spaceId);
 }
+async function createSpaceCollection(client2, spaceId) {
+  const collectionName = spaceId === "public" ? PUBLIC_COLLECTION_NAME : getSpaceCollectionName(spaceId);
+  logger.info("Creating space collection", {
+    module: "weaviate-space-schema",
+    collectionName,
+    spaceId
+  });
+  await client2.collections.create({
+    name: collectionName,
+    // Vectorizer configuration
+    vectorizers: weaviate3.configure.vectorizer.text2VecOpenAI({
+      model: "text-embedding-3-small",
+      // Vectorize content, title, summary, and observation for semantic search
+      // Note: title and summary are optional fields
+      sourceProperties: ["content", "title", "summary", "observation"]
+    }),
+    properties: [
+      // Discriminator
+      {
+        name: "doc_type",
+        dataType: "text",
+        description: 'Document type: "space_memory"'
+      },
+      // Space identity
+      {
+        name: "spaces",
+        dataType: "text[]",
+        description: 'Spaces this memory is published to (e.g., ["the_void", "dogs"])'
+      },
+      {
+        name: "space_id",
+        dataType: "text",
+        description: "DEPRECATED: Use spaces array instead. Will be removed in v3.0.0."
+      },
+      {
+        name: "author_id",
+        dataType: "text",
+        description: "Original author user_id (for permissions)"
+      },
+      {
+        name: "ghost_id",
+        dataType: "text",
+        description: "Optional ghost profile ID for pseudonymous publishing"
+      },
+      {
+        name: "attribution",
+        dataType: "text",
+        description: 'Attribution type: "user" or "ghost"'
+      },
+      // Discovery metadata
+      {
+        name: "published_at",
+        dataType: "text",
+        description: "When published to space (ISO 8601)"
+      },
+      {
+        name: "discovery_count",
+        dataType: "number",
+        description: "How many times discovered"
+      },
+      // Memory fields (same as personal memories)
+      {
+        name: "content",
+        dataType: "text",
+        description: "Main memory content (vectorized)"
+      },
+      {
+        name: "title",
+        dataType: "text",
+        description: "Optional short title"
+      },
+      {
+        name: "summary",
+        dataType: "text",
+        description: "Optional brief summary"
+      },
+      {
+        name: "type",
+        dataType: "text",
+        description: "Content type (note, event, person, etc.)"
+      },
+      // Scoring fields
+      {
+        name: "weight",
+        dataType: "number",
+        description: "Significance/priority (0-1)"
+      },
+      {
+        name: "trust",
+        dataType: "number",
+        description: "Access control level (0-1)"
+      },
+      {
+        name: "confidence",
+        dataType: "number",
+        description: "System confidence in accuracy (0-1)"
+      },
+      // Location fields (flattened) - MUST match user memory schema exactly
+      {
+        name: "location_gps_lat",
+        dataType: "number",
+        description: "GPS latitude"
+      },
+      {
+        name: "location_gps_lng",
+        dataType: "number",
+        description: "GPS longitude"
+      },
+      {
+        name: "location_address",
+        dataType: "text",
+        description: "Formatted address"
+      },
+      {
+        name: "location_city",
+        dataType: "text",
+        description: "City name"
+      },
+      {
+        name: "location_country",
+        dataType: "text",
+        description: "Country name"
+      },
+      {
+        name: "location_source",
+        dataType: "text",
+        description: "Location source (gps, ip, manual, etc.)"
+      },
+      // Context fields (flattened) - MUST match user memory schema exactly
+      {
+        name: "context_conversation_id",
+        dataType: "text",
+        description: "Conversation ID"
+      },
+      {
+        name: "context_summary",
+        dataType: "text",
+        description: "Brief context summary"
+      },
+      {
+        name: "context_timestamp",
+        dataType: "date",
+        description: "Context timestamp"
+      },
+      // Locale fields - MUST match user memory schema exactly
+      {
+        name: "locale_language",
+        dataType: "text",
+        description: "Language code (e.g., en, es, fr)"
+      },
+      {
+        name: "locale_timezone",
+        dataType: "text",
+        description: "Timezone (e.g., America/Los_Angeles)"
+      },
+      // Relationships - MUST match user memory schema exactly
+      {
+        name: "relationships",
+        dataType: "text[]",
+        description: "Array of relationship IDs"
+      },
+      // Access tracking - MUST match user memory schema exactly
+      {
+        name: "access_count",
+        dataType: "number",
+        description: "Total times accessed"
+      },
+      {
+        name: "last_accessed_at",
+        dataType: "date",
+        description: "Most recent access timestamp"
+      },
+      // Metadata - MUST match user memory schema exactly
+      {
+        name: "tags",
+        dataType: "text[]",
+        description: "Tags for organization"
+      },
+      {
+        name: "references",
+        dataType: "text[]",
+        description: "Source URLs"
+      },
+      {
+        name: "template_id",
+        dataType: "text",
+        description: "Template ID if using template"
+      },
+      // Relationship-specific fields (for relationships in public space)
+      {
+        name: "memory_ids",
+        dataType: "text[]",
+        description: "Connected memory IDs (for relationships)"
+      },
+      {
+        name: "relationship_type",
+        dataType: "text",
+        description: "Relationship type (for relationships)"
+      },
+      {
+        name: "observation",
+        dataType: "text",
+        description: "Relationship observation (vectorized)"
+      },
+      {
+        name: "strength",
+        dataType: "number",
+        description: "Relationship strength (0-1)"
+      },
+      // Computed fields - MUST match user memory schema exactly
+      {
+        name: "base_weight",
+        dataType: "number",
+        description: "User-specified weight"
+      },
+      {
+        name: "computed_weight",
+        dataType: "number",
+        description: "Calculated effective weight"
+      },
+      // User ID field (for backwards compatibility with spread operator)
+      {
+        name: "user_id",
+        dataType: "text",
+        description: "User ID (copied from original memory, same as author_id)"
+      },
+      // Timestamps
+      {
+        name: "created_at",
+        dataType: "text",
+        description: "Original creation timestamp (ISO 8601)"
+      },
+      {
+        name: "updated_at",
+        dataType: "text",
+        description: "Last update timestamp (ISO 8601)"
+      },
+      // Versioning
+      {
+        name: "version",
+        dataType: "number",
+        description: "Version number (increments on update)"
+      },
+      // Comment/threading fields (for threaded discussions in shared spaces)
+      {
+        name: "parent_id",
+        dataType: "text",
+        description: "ID of parent memory or comment (for threading)"
+      },
+      {
+        name: "thread_root_id",
+        dataType: "text",
+        description: "Root memory ID for fetching entire thread"
+      },
+      {
+        name: "moderation_flags",
+        dataType: "text[]",
+        description: 'Per-space moderation flags (format: "{space_id}:{flag_type}")'
+      }
+    ]
+  });
+  logger.info("Space collection created successfully", {
+    module: "weaviate-space-schema",
+    collectionName
+  });
+}
 async function ensurePublicCollection(client2) {
   const collectionName = PUBLIC_COLLECTION_NAME;
+  const exists = await client2.collections.exists(collectionName);
+  if (!exists) {
+    await createSpaceCollection(client2, "public");
+  }
   return client2.collections.get(collectionName);
 }
 
@@ -4057,7 +4330,7 @@ async function executePublishMemory(request, userId) {
       attribution: "user",
       // Merge additional tags with original tags
       tags: [...originalTags, ...additionalTags]
-      // Keep doc_type as 'memory' (don't change to 'space_memory')
+      // Keep doc_type as 'memory' (space_memory concept was removed)
       // Keep original created_at, updated_at, version (don't overwrite)
     };
     logger.info("Inserting memory into Memory_public", {
@@ -4269,7 +4542,7 @@ async function handleSearchSpace(args, userId) {
     const publicCollection = await ensurePublicCollection(weaviateClient);
     const filterList = [];
     filterList.push(publicCollection.filter.byProperty("spaces").containsAny(args.spaces));
-    filterList.push(publicCollection.filter.byProperty("doc_type").equal("space_memory"));
+    filterList.push(publicCollection.filter.byProperty("doc_type").equal("memory"));
     if (args.content_type) {
       filterList.push(publicCollection.filter.byProperty("type").equal(args.content_type));
     }
@@ -4425,7 +4698,7 @@ async function handleQuerySpace(args, userId) {
     const publicCollection = await ensurePublicCollection(weaviateClient);
     const filterList = [];
     filterList.push(publicCollection.filter.byProperty("spaces").containsAny(args.spaces));
-    filterList.push(publicCollection.filter.byProperty("doc_type").equal("space_memory"));
+    filterList.push(publicCollection.filter.byProperty("doc_type").equal("memory"));
     if (args.content_type) {
       filterList.push(publicCollection.filter.byProperty("type").equal(args.content_type));
     }

package/dist/server.js

@@ -3756,11 +3756,284 @@ init_space_memory();
 init_logger();
 import weaviate3 from "weaviate-client";
 var PUBLIC_COLLECTION_NAME = "Memory_public";
+function getSpaceCollectionName(spaceId) {
+  return `Memory_${spaceId}`;
+}
 function isValidSpaceId(spaceId) {
   return SUPPORTED_SPACES.includes(spaceId);
 }
+async function createSpaceCollection(client2, spaceId) {
+  const collectionName = spaceId === "public" ? PUBLIC_COLLECTION_NAME : getSpaceCollectionName(spaceId);
+  logger.info("Creating space collection", {
+    module: "weaviate-space-schema",
+    collectionName,
+    spaceId
+  });
+  await client2.collections.create({
+    name: collectionName,
+    // Vectorizer configuration
+    vectorizers: weaviate3.configure.vectorizer.text2VecOpenAI({
+      model: "text-embedding-3-small",
+      // Vectorize content, title, summary, and observation for semantic search
+      // Note: title and summary are optional fields
+      sourceProperties: ["content", "title", "summary", "observation"]
+    }),
+    properties: [
+      // Discriminator
+      {
+        name: "doc_type",
+        dataType: "text",
+        description: 'Document type: "space_memory"'
+      },
+      // Space identity
+      {
+        name: "spaces",
+        dataType: "text[]",
+        description: 'Spaces this memory is published to (e.g., ["the_void", "dogs"])'
+      },
+      {
+        name: "space_id",
+        dataType: "text",
+        description: "DEPRECATED: Use spaces array instead. Will be removed in v3.0.0."
+      },
+      {
+        name: "author_id",
+        dataType: "text",
+        description: "Original author user_id (for permissions)"
+      },
+      {
+        name: "ghost_id",
+        dataType: "text",
+        description: "Optional ghost profile ID for pseudonymous publishing"
+      },
+      {
+        name: "attribution",
+        dataType: "text",
+        description: 'Attribution type: "user" or "ghost"'
+      },
+      // Discovery metadata
+      {
+        name: "published_at",
+        dataType: "text",
+        description: "When published to space (ISO 8601)"
+      },
+      {
+        name: "discovery_count",
+        dataType: "number",
+        description: "How many times discovered"
+      },
+      // Memory fields (same as personal memories)
+      {
+        name: "content",
+        dataType: "text",
+        description: "Main memory content (vectorized)"
+      },
+      {
+        name: "title",
+        dataType: "text",
+        description: "Optional short title"
+      },
+      {
+        name: "summary",
+        dataType: "text",
+        description: "Optional brief summary"
+      },
+      {
+        name: "type",
+        dataType: "text",
+        description: "Content type (note, event, person, etc.)"
+      },
+      // Scoring fields
+      {
+        name: "weight",
+        dataType: "number",
+        description: "Significance/priority (0-1)"
+      },
+      {
+        name: "trust",
+        dataType: "number",
+        description: "Access control level (0-1)"
+      },
+      {
+        name: "confidence",
+        dataType: "number",
+        description: "System confidence in accuracy (0-1)"
+      },
+      // Location fields (flattened) - MUST match user memory schema exactly
+      {
+        name: "location_gps_lat",
+        dataType: "number",
+        description: "GPS latitude"
+      },
+      {
+        name: "location_gps_lng",
+        dataType: "number",
+        description: "GPS longitude"
+      },
+      {
+        name: "location_address",
+        dataType: "text",
+        description: "Formatted address"
+      },
+      {
+        name: "location_city",
+        dataType: "text",
+        description: "City name"
+      },
+      {
+        name: "location_country",
+        dataType: "text",
+        description: "Country name"
+      },
+      {
+        name: "location_source",
+        dataType: "text",
+        description: "Location source (gps, ip, manual, etc.)"
+      },
+      // Context fields (flattened) - MUST match user memory schema exactly
+      {
+        name: "context_conversation_id",
+        dataType: "text",
+        description: "Conversation ID"
+      },
+      {
+        name: "context_summary",
+        dataType: "text",
+        description: "Brief context summary"
+      },
+      {
+        name: "context_timestamp",
+        dataType: "date",
+        description: "Context timestamp"
+      },
+      // Locale fields - MUST match user memory schema exactly
+      {
+        name: "locale_language",
+        dataType: "text",
+        description: "Language code (e.g., en, es, fr)"
+      },
+      {
+        name: "locale_timezone",
+        dataType: "text",
+        description: "Timezone (e.g., America/Los_Angeles)"
+      },
+      // Relationships - MUST match user memory schema exactly
+      {
+        name: "relationships",
+        dataType: "text[]",
+        description: "Array of relationship IDs"
+      },
+      // Access tracking - MUST match user memory schema exactly
+      {
+        name: "access_count",
+        dataType: "number",
+        description: "Total times accessed"
+      },
+      {
+        name: "last_accessed_at",
+        dataType: "date",
+        description: "Most recent access timestamp"
+      },
+      // Metadata - MUST match user memory schema exactly
+      {
+        name: "tags",
+        dataType: "text[]",
+        description: "Tags for organization"
+      },
+      {
+        name: "references",
+        dataType: "text[]",
+        description: "Source URLs"
+      },
+      {
+        name: "template_id",
+        dataType: "text",
+        description: "Template ID if using template"
+      },
+      // Relationship-specific fields (for relationships in public space)
+      {
+        name: "memory_ids",
+        dataType: "text[]",
+        description: "Connected memory IDs (for relationships)"
+      },
+      {
+        name: "relationship_type",
+        dataType: "text",
+        description: "Relationship type (for relationships)"
+      },
+      {
+        name: "observation",
+        dataType: "text",
+        description: "Relationship observation (vectorized)"
+      },
+      {
+        name: "strength",
+        dataType: "number",
+        description: "Relationship strength (0-1)"
+      },
+      // Computed fields - MUST match user memory schema exactly
+      {
+        name: "base_weight",
+        dataType: "number",
+        description: "User-specified weight"
+      },
+      {
+        name: "computed_weight",
+        dataType: "number",
+        description: "Calculated effective weight"
+      },
+      // User ID field (for backwards compatibility with spread operator)
+      {
+        name: "user_id",
+        dataType: "text",
+        description: "User ID (copied from original memory, same as author_id)"
+      },
+      // Timestamps
+      {
+        name: "created_at",
+        dataType: "text",
+        description: "Original creation timestamp (ISO 8601)"
+      },
+      {
+        name: "updated_at",
+        dataType: "text",
+        description: "Last update timestamp (ISO 8601)"
+      },
+      // Versioning
+      {
+        name: "version",
+        dataType: "number",
+        description: "Version number (increments on update)"
+      },
+      // Comment/threading fields (for threaded discussions in shared spaces)
+      {
+        name: "parent_id",
+        dataType: "text",
+        description: "ID of parent memory or comment (for threading)"
+      },
+      {
+        name: "thread_root_id",
+        dataType: "text",
+        description: "Root memory ID for fetching entire thread"
+      },
+      {
+        name: "moderation_flags",
+        dataType: "text[]",
+        description: 'Per-space moderation flags (format: "{space_id}:{flag_type}")'
+      }
+    ]
+  });
+  logger.info("Space collection created successfully", {
+    module: "weaviate-space-schema",
+    collectionName
+  });
+}
 async function ensurePublicCollection(client2) {
   const collectionName = PUBLIC_COLLECTION_NAME;
+  const exists = await client2.collections.exists(collectionName);
+  if (!exists) {
+    await createSpaceCollection(client2, "public");
+  }
   return client2.collections.get(collectionName);
 }
 
@@ -4125,7 +4398,7 @@ async function executePublishMemory(request, userId) {
       attribution: "user",
       // Merge additional tags with original tags
       tags: [...originalTags, ...additionalTags]
-      // Keep doc_type as 'memory' (don't change to 'space_memory')
+      // Keep doc_type as 'memory' (space_memory concept was removed)
       // Keep original created_at, updated_at, version (don't overwrite)
     };
     logger.info("Inserting memory into Memory_public", {
@@ -4337,7 +4610,7 @@ async function handleSearchSpace(args, userId) {
     const publicCollection = await ensurePublicCollection(weaviateClient);
     const filterList = [];
     filterList.push(publicCollection.filter.byProperty("spaces").containsAny(args.spaces));
-    filterList.push(publicCollection.filter.byProperty("doc_type").equal("space_memory"));
+    filterList.push(publicCollection.filter.byProperty("doc_type").equal("memory"));
     if (args.content_type) {
       filterList.push(publicCollection.filter.byProperty("type").equal(args.content_type));
     }
@@ -4493,7 +4766,7 @@ async function handleQuerySpace(args, userId) {
     const publicCollection = await ensurePublicCollection(weaviateClient);
     const filterList = [];
     filterList.push(publicCollection.filter.byProperty("spaces").containsAny(args.spaces));
-    filterList.push(publicCollection.filter.byProperty("doc_type").equal("space_memory"));
+    filterList.push(publicCollection.filter.byProperty("doc_type").equal("memory"));
     if (args.content_type) {
       filterList.push(publicCollection.filter.byProperty("type").equal(args.content_type));
     }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@prmichaelsen/remember-mcp",
-  "version": "2.7.7",
+  "version": "2.7.9",
   "description": "Multi-tenant memory system MCP server with vector search and relationships",
   "main": "dist/server.js",
   "type": "module",

package/src/tools/confirm.ts

@@ -243,7 +243,7 @@ async function executePublishMemory(
       attribution: 'user' as const,
       // Merge additional tags with original tags
       tags: [...originalTags, ...additionalTags],
-      // Keep doc_type as 'memory' (don't change to 'space_memory')
+      // Keep doc_type as 'memory' (space_memory concept was removed)
       // Keep original created_at, updated_at, version (don't overwrite)
     };
 

package/src/tools/query-space.ts

@@ -143,8 +143,8 @@ export async function handleQuerySpace(
     // Filter by spaces array (memory must be in at least one requested space)
     filterList.push(publicCollection.filter.byProperty('spaces').containsAny(args.spaces));
 
-    // Filter by doc_type (space_memory)
-    filterList.push(publicCollection.filter.byProperty('doc_type').equal('space_memory'));
+    // Filter by doc_type (memory) - space_memory concept was removed
+    filterList.push(publicCollection.filter.byProperty('doc_type').equal('memory'));
 
     // Apply content type filter
     if (args.content_type) {

package/src/tools/search-space.ts

@@ -155,8 +155,8 @@ export async function handleSearchSpace(
     // Filter by spaces array (memory must be in at least one requested space)
     filterList.push(publicCollection.filter.byProperty('spaces').containsAny(args.spaces));
 
-    // Filter by doc_type (space_memory)
-    filterList.push(publicCollection.filter.byProperty('doc_type').equal('space_memory'));
+    // Filter by doc_type (memory) - space_memory concept was removed
+    filterList.push(publicCollection.filter.byProperty('doc_type').equal('memory'));
 
     // Apply content type filter
     if (args.content_type) {

package/src/weaviate/space-schema.ts

@@ -384,16 +384,14 @@ export async function ensurePublicCollection(
 ): Promise<Collection<any>> {
   const collectionName = PUBLIC_COLLECTION_NAME;
 
-  // DON'T manually create the collection!
-  // Let Weaviate auto-generate the schema from the first insert.
-  // This ensures the schema matches user memory collections exactly,
-  // including nested objects (context, location) that were auto-generated.
-  //
-  // When we spread {...originalMemory.properties}, the nested objects
-  // will be preserved and Weaviate will auto-create matching schema.
-  //
-  // NOTE: Collection will be created automatically on first insert.
-  // Weaviate's auto-schema feature handles this transparently.
+  // Check if collection exists
+  const exists = await client.collections.exists(collectionName);
+  
+  if (!exists) {
+    // Create with proper vectorizer configuration
+    // Use 'public' as spaceId to trigger PUBLIC_COLLECTION_NAME
+    await createSpaceCollection(client, 'public');
+  }
 
   return client.collections.get(collectionName);
 }