[ [0x60] Features ] AI Memory

AI Memory

> Multi-scope vector memory for contextual AI conversations.

Overview

[ [0x60] STATUS ]

The AI Memory system provides persistent, semantic memory for your AI features. Memories are stored with vector embeddings and retrieved via similarity search. Supports multiple scopes (chat, project, user) so context stays relevant. Uses Qdrant for production and an in-memory fallback for development — gracefully degrades when services are unavailable.

Features

[ [0x01] SEMANTIC_SEA ]

STATUS: READY

DESC: Find relevant memories by meaning, not just keywords. Uses vector embeddings for similarity matching.

[ [0x02] MULTI_SCOPE_ ]

STATUS: READY

DESC: Store memories at chat, project, or user scope. Each scope is isolated for relevance.

[ [0x03] QDRANT_VECTO ]

STATUS: READY

DESC: Production-ready Qdrant integration via HTTP API. No SDK dependency required.

[ [0x04] GRACEFUL_DEG ]

STATUS: READY

DESC: Falls back to in-memory store when Qdrant is unavailable. AI features keep working.

[ [0x05] MEMORY_API ]

STATUS: READY

DESC: REST API for adding, searching, and deleting memories. Auth-protected with session validation.

[ [0x06] AUTO_EMBEDDI ]

STATUS: READY

DESC: Embeddings generated automatically using the configured AI provider. Supports OpenAI and Google models.

Setup

[ [0x01] ENABLE VECTOR DATABASE ]

DESC: Set SERVICE_VECTOR_DB=true in your environment to use Qdrant for production memory storage.

1$# .env.local
2$SERVICE_VECTOR_DB=true
3$FEATURE_MEMORY=true
4
5$# Qdrant settings (defaults shown)
6$QDRANT_HOST=localhost
7$QDRANT_PORT=6333
8$QDRANT_COLLECTION=fabrk_memories

[ [0x02] START QDRANT (DOCKER) ]

DESC: Run Qdrant locally via Docker, or uncomment the qdrant service in docker-compose.yml.

1$docker run -p 6333:6333 qdrant/qdrant

[ [0x03] USE THE MEMORY API ]

DESC: Add and search memories via the REST API or directly in server code.

1// Server-side usage
2import { addMemory, searchMemory } from '@/lib/ai/memory';
3
4// Store a memory
5await addMemory('User prefers dark mode', 'user', userId);
6
7// Search memories
8const results = await searchMemory(
9  'theme preferences',
10  'user',
11  userId,
12  5 // limit
13);
14
15// REST API
16// POST /api/ai/memory { content, scope, scopeId }
17// GET  /api/ai/memory?query=...&scope=user&scopeId=...
18// DELETE /api/ai/memory { ids: [...] }

Overview

[ [0x60] STATUS ]

Features

[ [0x01] SEMANTIC_SEA ]

STATUS: READY

DESC: Find relevant memories by meaning, not just keywords. Uses vector embeddings for similarity matching.

[ [0x02] MULTI_SCOPE_ ]

STATUS: READY

DESC: Store memories at chat, project, or user scope. Each scope is isolated for relevance.

[ [0x03] QDRANT_VECTO ]

STATUS: READY

DESC: Production-ready Qdrant integration via HTTP API. No SDK dependency required.

[ [0x04] GRACEFUL_DEG ]

STATUS: READY

DESC: Falls back to in-memory store when Qdrant is unavailable. AI features keep working.

[ [0x05] MEMORY_API ]

STATUS: READY

DESC: REST API for adding, searching, and deleting memories. Auth-protected with session validation.

[ [0x06] AUTO_EMBEDDI ]

STATUS: READY

DESC: Embeddings generated automatically using the configured AI provider. Supports OpenAI and Google models.

Setup

[ [0x01] ENABLE VECTOR DATABASE ]

DESC: Set SERVICE_VECTOR_DB=true in your environment to use Qdrant for production memory storage.

1$# .env.local
2$SERVICE_VECTOR_DB=true
3$FEATURE_MEMORY=true
4
5$# Qdrant settings (defaults shown)
6$QDRANT_HOST=localhost
7$QDRANT_PORT=6333
8$QDRANT_COLLECTION=fabrk_memories

[ [0x02] START QDRANT (DOCKER) ]

DESC: Run Qdrant locally via Docker, or uncomment the qdrant service in docker-compose.yml.

1$docker run -p 6333:6333 qdrant/qdrant

[ [0x03] USE THE MEMORY API ]

DESC: Add and search memories via the REST API or directly in server code.

1// Server-side usage
2import { addMemory, searchMemory } from '@/lib/ai/memory';
3
4// Store a memory
5await addMemory('User prefers dark mode', 'user', userId);
6
7// Search memories
8const results = await searchMemory(
9  'theme preferences',
10  'user',
11  userId,
12  5 // limit
13);
14
15// REST API
16// POST /api/ai/memory { content, scope, scopeId }
17// GET  /api/ai/memory?query=...&scope=user&scopeId=...
18// DELETE /api/ai/memory { ids: [...] }