POST /memory/search Search memories based on a query using semantic similarity. Returns relevant memories with similarity scores. Optionally filter results by memory type.
Request Body The search query. The API will find memories semantically similar to this query.
Unique identifier for the user whose memories to search.
Maximum number of results to return. Defaults to 10.
Filter results by memory type. Accepts: LONG_TERM, SHORT_TERM, EPISODIC, SEMANTIC, or PROCEDURAL.
Response Indicates whether the operation was successful.
Array of matching memories sorted by relevance. Show Memory Result Object
Unique identifier for the memory.
The extracted memory content.
Similarity score indicating relevance to the query (higher is better).
Associated metadata including memory_type and custom fields.
The user ID this memory belongs to.
ISO 8601 timestamp of when the memory was created.
ISO 8601 timestamp of when the memory was last updated.
Example Request curl -X POST http://localhost:8000/memory/search \ -H "Content-Type: application/json" \ -d '{ "query": "What are my UI preferences?", "user_id": "user_123", "limit": 5 }'
Example Response { "success" : true , "results" : [ { "id" : "mem_abc123" , "memory" : "User prefers dark mode in applications" , "score" : 0.92 , "metadata" : { "memory_type" : "LONG_TERM" }, "user_id" : "user_123" , "created_at" : "2026-03-01T10:30:00Z" , "updated_at" : "2026-03-01T10:30:00Z" }, { "id" : "mem_def456" , "memory" : "User likes minimal UI designs with high contrast" , "score" : 0.87 , "metadata" : { "memory_type" : "LONG_TERM" }, "user_id" : "user_123" , "created_at" : "2026-02-28T14:20:00Z" , "updated_at" : "2026-02-28T14:20:00Z" } ] }
Filtering by Memory Type Search only for episodic memories (past events):
curl -X POST http://localhost:8000/memory/search \ -H "Content-Type: application/json" \ -d '{ "query": "conferences I attended", "user_id": "user_123", "memory_type": "EPISODIC", "limit": 10 }'
Search Short-Term Memories Find what the user is currently working on:
curl -X POST http://localhost:8000/memory/search \ -H "Content-Type: application/json" \ -d '{ "query": "current tasks", "user_id": "user_123", "memory_type": "SHORT_TERM" }'
Error Responses Error message describing what went wrong.
500 Internal Server Error { "detail" : "Failed to execute search query" }
Implementation Details
Source: backend/main.py:221Search Method: Semantic similarity using vector embeddingsIndex: Supabase HNSW index with cosine distanceFiltering: Supports metadata-based filtering for memory_type and custom fields