@prmichaelsen/remember-mcp 2.2.1 → 2.3.1

package/agent/tasks/task-38-implement-remember-deny.md

@@ -0,0 +1,161 @@
+# Task 38: Implement remember_deny Tool
+
+**Milestone**: M10 - Shared Spaces & Confirmation Flow
+**Estimated Time**: 2 hours
+**Dependencies**: Task 34 (Token Service)
+**Status**: Not Started
+
+---
+
+## Objective
+
+Implement the generic `remember_deny` tool that cancels any pending action. This allows users to reject confirmation requests.
+
+---
+
+## Steps
+
+### 1. Create Tool File
+
+Create `src/tools/deny.ts` with denial logic.
+
+**Actions**:
+- Import token service
+- Define tool schema
+- Create handler function
+- Export tool definition and handler
+
+**Expected Outcome**: Tool file structure created
+
+### 2. Define Tool Schema
+
+Create MCP tool definition for remember_deny.
+
+**Actions**:
+- Set tool name: `remember_deny`
+- Write clear description emphasizing generic nature
+- Define input schema with single parameter:
+  - `token` (required): Confirmation token to deny
+- Add helpful description
+
+**Expected Outcome**: Tool schema complete
+
+### 3. Implement Token Denial
+
+Use token service to deny the request.
+
+**Actions**:
+- Call `confirmationTokenService.denyRequest()`
+- Pass userId and token
+- Check if denial was successful
+- Handle invalid token error
+- Return appropriate response
+
+**Expected Outcome**: Token denial working
+
+### 4. Format Success Response
+
+Return simple success response.
+
+**Actions**:
+- Create response with success flag
+- Keep response minimal (just confirmation of denial)
+- Format as JSON string
+- No payload needed for denial
+
+**Expected Outcome**: Clear response format
+
+### 5. Implement Error Handling
+
+Handle error cases.
+
+**Actions**:
+- Invalid token error
+- Token already used error
+- Token not found error
+- Use `handleToolError` utility
+- Include context in errors
+
+**Expected Outcome**: Error handling complete
+
+### 6. Add Tool to Server
+
+Register tool in both server files.
+
+**Actions**:
+- Import tool in `src/server.ts`
+- Add to tools list
+- Add to call handler
+- Repeat for `src/server-factory.ts`
+
+**Expected Outcome**: Tool available in MCP server
+
+### 7. Create Unit Tests
+
+Test the deny tool.
+
+**Actions**:
+- Create `tests/unit/deny.test.ts`
+- Test successful denial
+- Test invalid token error
+- Test token already used error
+- Mock token service
+- Verify response format
+
+**Expected Outcome**: All tests passing
+
+---
+
+## Verification
+
+- [ ] `src/tools/deny.ts` created
+- [ ] Tool schema defined
+- [ ] Token denial implemented
+- [ ] Success response formatted correctly
+- [ ] Error handling complete
+- [ ] Tool registered in server.ts
+- [ ] Tool registered in server-factory.ts
+- [ ] Unit tests created and passing
+- [ ] TypeScript compiles without errors
+
+---
+
+## Tool Schema
+
+```typescript
+export const denyTool = {
+  name: 'remember_deny',
+  description: 'Deny a pending action. The request will be marked as denied and the token invalidated. Works for any action that requires confirmation.',
+  inputSchema: {
+    type: 'object',
+    properties: {
+      token: {
+        type: 'string',
+        description: 'The confirmation token from the action tool'
+      }
+    },
+    required: ['token']
+  }
+};
+```
+
+---
+
+## Response Format
+
+```json
+{
+  "success": true
+}
+```
+
+---
+
+## Related Files
+
+- Design: [`agent/design/publish-tools-confirmation-flow.md`](../design/publish-tools-confirmation-flow.md)
+- Token Service: [`src/services/confirmation-token.service.ts`](../../src/services/confirmation-token.service.ts)
+
+---
+
+**Next Task**: Task 39 - Implement remember_search_space Tool

package/agent/tasks/task-39-implement-remember-search-space.md

@@ -0,0 +1,188 @@
+# Task 39: Implement remember_search_space Tool
+
+**Milestone**: M10 - Shared Spaces & Confirmation Flow
+**Estimated Time**: 3 hours
+**Dependencies**: Task 35 (Space Types), M2 (Search Memory tool)
+**Status**: Not Started
+
+---
+
+## Objective
+
+Implement `remember_search_space` tool for searching shared spaces. This is similar to `remember_search_memory` but searches space collections instead of personal collections.
+
+---
+
+## Steps
+
+### 1. Create Tool File
+
+Create `src/tools/search-space.ts` based on search-memory.ts.
+
+**Actions**:
+- Copy structure from `search-memory.ts`
+- Import space schema utilities
+- Define tool schema
+- Create handler function
+- Export tool definition and handler
+
+**Expected Outcome**: Tool file structure created
+
+### 2. Define Tool Schema
+
+Create MCP tool definition for remember_search_space.
+
+**Actions**:
+- Set tool name: `remember_search_space`
+- Write clear description for searching shared spaces
+- Define input schema with properties:
+  - `query` (required): Search query
+  - `space` (required): Space ID (enum: ['the_void'])
+  - All filter options from search_memory (content_type, tags, weight, trust, dates)
+  - `limit` and `offset` for pagination
+- Set default values
+
+**Expected Outcome**: Tool schema complete
+
+### 3. Implement Space Collection Access
+
+Get the space collection to search.
+
+**Actions**:
+- Call `ensureSpaceCollection()` with space_id
+- Get collection reference
+- Handle collection creation if needed
+- Verify collection exists
+
+**Expected Outcome**: Space collection accessible
+
+### 4. Implement Hybrid Search
+
+Execute hybrid search on space collection.
+
+**Actions**:
+- Use Weaviate hybrid search (semantic + keyword)
+- Apply all filters (content_type, tags, weight, trust, dates)
+- Filter by `space_id` (not `user_id`)
+- Use same filter building logic as search_memory
+- Handle pagination with offset/limit
+- Return results with scores
+
+**Expected Outcome**: Search working correctly
+
+### 5. Format Search Results
+
+Format results for agent consumption.
+
+**Actions**:
+- Extract memory properties
+- Include relevance scores
+- Format as array of SpaceMemory objects
+- Include metadata (total results, offset, limit)
+- Format as JSON string
+
+**Expected Outcome**: Clear result format
+
+### 6. Implement Error Handling
+
+Handle error cases.
+
+**Actions**:
+- Invalid space ID error
+- Collection not found error
+- Weaviate query errors
+- Use `handleToolError` utility
+- Include context in errors
+
+**Expected Outcome**: Error handling complete
+
+### 7. Add Tool to Server
+
+Register tool in both server files.
+
+**Actions**:
+- Import tool in `src/server.ts`
+- Add to tools list
+- Add to call handler
+- Repeat for `src/server-factory.ts`
+
+**Expected Outcome**: Tool available in MCP server
+
+### 8. Create Unit Tests
+
+Test the search_space tool.
+
+**Actions**:
+- Create `tests/unit/search-space.test.ts`
+- Test successful search
+- Test with various filters
+- Test pagination
+- Test invalid space ID error
+- Mock Weaviate client
+- Verify result format
+
+**Expected Outcome**: All tests passing
+
+---
+
+## Verification
+
+- [ ] `src/tools/search-space.ts` created
+- [ ] Tool schema defined with all parameters
+- [ ] Space collection access working
+- [ ] Hybrid search implemented
+- [ ] Filters working correctly
+- [ ] Results formatted properly
+- [ ] Error handling complete
+- [ ] Tool registered in server.ts
+- [ ] Tool registered in server-factory.ts
+- [ ] Unit tests created and passing
+- [ ] TypeScript compiles without errors
+
+---
+
+## Tool Schema
+
+```typescript
+export const searchSpaceTool = {
+  name: 'remember_search_space',
+  description: 'Search shared spaces to discover thoughts, ideas, and memories. Works like remember_search_memory but searches shared spaces instead of personal memories.',
+  inputSchema: {
+    type: 'object',
+    properties: {
+      query: {
+        type: 'string',
+        description: 'Search query (semantic + keyword hybrid)'
+      },
+      space: {
+        type: 'string',
+        description: 'Which space to search',
+        enum: ['the_void'],
+        default: 'the_void'
+      },
+      // Same filters as remember_search_memory
+      content_type: { type: 'string' },
+      tags: { type: 'array', items: { type: 'string' } },
+      min_weight: { type: 'number', minimum: 0, maximum: 1 },
+      max_weight: { type: 'number', minimum: 0, maximum: 1 },
+      date_from: { type: 'string' },
+      date_to: { type: 'string' },
+      limit: { type: 'number', default: 10 },
+      offset: { type: 'number', default: 0 }
+    },
+    required: ['query', 'space']
+  }
+};
+```
+
+---
+
+## Related Files
+
+- Design: [`agent/design/publish-tools-confirmation-flow.md`](../design/publish-tools-confirmation-flow.md)
+- Search Memory: [`src/tools/search-memory.ts`](../../src/tools/search-memory.ts)
+- Space Schema: [`src/weaviate/space-schema.ts`](../../src/weaviate/space-schema.ts)
+
+---
+
+**Next Task**: Task 40 - Implement remember_query_space Tool

package/agent/tasks/task-40-implement-remember-query-space.md

@@ -0,0 +1,193 @@
+# Task 40: Implement remember_query_space Tool
+
+**Milestone**: M10 - Shared Spaces & Confirmation Flow
+**Estimated Time**: 3 hours
+**Dependencies**: Task 35 (Space Types), M2 (Query Memory tool)
+**Status**: Not Started
+
+---
+
+## Objective
+
+Implement `remember_query_space` tool for RAG-optimized natural language queries on shared spaces. Similar to `remember_query_memory` but queries space collections.
+
+---
+
+## Steps
+
+### 1. Create Tool File
+
+Create `src/tools/query-space.ts` based on query-memory.ts.
+
+**Actions**:
+- Copy structure from `query-memory.ts`
+- Import space schema utilities
+- Define tool schema
+- Create handler function
+- Export tool definition and handler
+
+**Expected Outcome**: Tool file structure created
+
+### 2. Define Tool Schema
+
+Create MCP tool definition for remember_query_space.
+
+**Actions**:
+- Set tool name: `remember_query_space`
+- Write clear description for RAG queries on shared spaces
+- Define input schema with properties:
+  - `question` (required): Natural language question
+  - `space` (required): Space ID (enum: ['the_void'])
+  - All filter options from query_memory
+  - `format`: 'detailed' or 'compact'
+  - `limit` for result count
+- Set default values
+
+**Expected Outcome**: Tool schema complete
+
+### 3. Implement Space Collection Access
+
+Get the space collection to query.
+
+**Actions**:
+- Call `ensureSpaceCollection()` with space_id
+- Get collection reference
+- Handle collection creation if needed
+- Verify collection exists
+
+**Expected Outcome**: Space collection accessible
+
+### 4. Implement RAG Query
+
+Execute semantic query on space collection.
+
+**Actions**:
+- Use Weaviate nearText for semantic search
+- Apply all filters (content_type, tags, weight, trust, dates)
+- Filter by `space_id` (not `user_id`)
+- Use same filter building logic as query_memory
+- Return results with relevance scores
+- Limit results appropriately
+
+**Expected Outcome**: RAG query working
+
+### 5. Format Query Results
+
+Format results based on requested format.
+
+**Actions**:
+- **Detailed format**: Full SpaceMemory objects with all fields
+- **Compact format**: Text summary optimized for LLM context
+- Include relevance scores
+- Include metadata (question, space, result count)
+- Format as JSON string
+
+**Expected Outcome**: Both formats working
+
+### 6. Implement Error Handling
+
+Handle error cases.
+
+**Actions**:
+- Invalid space ID error
+- Collection not found error
+- Weaviate query errors
+- Use `handleToolError` utility
+- Include context in errors
+
+**Expected Outcome**: Error handling complete
+
+### 7. Add Tool to Server
+
+Register tool in both server files.
+
+**Actions**:
+- Import tool in `src/server.ts`
+- Add to tools list
+- Add to call handler
+- Repeat for `src/server-factory.ts`
+
+**Expected Outcome**: Tool available in MCP server
+
+### 8. Create Unit Tests
+
+Test the query_space tool.
+
+**Actions**:
+- Create `tests/unit/query-space.test.ts`
+- Test successful query (detailed format)
+- Test successful query (compact format)
+- Test with various filters
+- Test invalid space ID error
+- Mock Weaviate client
+- Verify result formats
+
+**Expected Outcome**: All tests passing
+
+---
+
+## Verification
+
+- [ ] `src/tools/query-space.ts` created
+- [ ] Tool schema defined with all parameters
+- [ ] Space collection access working
+- [ ] RAG query implemented
+- [ ] Filters working correctly
+- [ ] Detailed format working
+- [ ] Compact format working
+- [ ] Error handling complete
+- [ ] Tool registered in server.ts
+- [ ] Tool registered in server-factory.ts
+- [ ] Unit tests created and passing
+- [ ] TypeScript compiles without errors
+
+---
+
+## Tool Schema
+
+```typescript
+export const querySpaceTool = {
+  name: 'remember_query_space',
+  description: 'Ask natural language questions about memories in shared spaces. Works like remember_query_memory but queries shared spaces.',
+  inputSchema: {
+    type: 'object',
+    properties: {
+      question: {
+        type: 'string',
+        description: 'Natural language question'
+      },
+      space: {
+        type: 'string',
+        description: 'Which space to query',
+        enum: ['the_void'],
+        default: 'the_void'
+      },
+      // Same filters as remember_query_memory
+      content_type: { type: 'string' },
+      tags: { type: 'array', items: { type: 'string' } },
+      min_weight: { type: 'number', minimum: 0, maximum: 1 },
+      date_from: { type: 'string' },
+      date_to: { type: 'string' },
+      limit: { type: 'number', default: 10 },
+      format: { 
+        type: 'string', 
+        enum: ['detailed', 'compact'], 
+        default: 'detailed' 
+      }
+    },
+    required: ['question', 'space']
+  }
+};
+```
+
+---
+
+## Related Files
+
+- Design: [`agent/design/publish-tools-confirmation-flow.md`](../design/publish-tools-confirmation-flow.md)
+- Query Memory: [`src/tools/query-memory.ts`](../../src/tools/query-memory.ts)
+- Space Schema: [`src/weaviate/space-schema.ts`](../../src/weaviate/space-schema.ts)
+
+---
+
+**Next Task**: Task 41 - Configure Firestore TTL Policy