Get Memory

Retrieves relevant stored memories for a specific user. You can fetch memories by providing either a direct search string (query) or an array of conversation messages (messages) to extract context automatically.

HTTP Request

POST /get_memory

Request Body

The request requires a JSON body containing the user identifier and the retrieval parameters.

Important: You must provide exactly one of either query or messages. Providing both or neither will result in a 400 Bad Request error (and a ValueError in the Python SDK).

Parameter	Type	Required	Description
userId	string	Yes	The unique identifier for the user whose memories are being retrieved.
query	string	No*	A direct text query used to find semantically similar memories.
messages	array	No*	A list of message objects representing the conversation history. Must contain at least one message if provided.
threshold	float	No	The similarity threshold for memory retrieval. Defaults to 0.65 if not provided.
core	bool	No	Decides if core memories are to be sent to the client. True by default unless false is passed.

Message Object Structure: If using the messages array, each item must be formatted as follows:

• role (string): The author of the message (e.g., "user", "model").
• content (string): The text body of the message.

Example Request Payload:

{
  "userId": "user_123",
  "threshold": 0.7,
  "messages": [
    {
      "role": "user",
      "content": "Need to buy a new cycle" 
    },
    {
      "role": "model",
      "content": "That sounds exciting! Do you have an itinerary yet?"
    }
  ],
}

---

Response

Returns an array of memory objects that meet the semantic threshold criteria.

Success Response

• Status Code: 200 OK
• Content-Type: application/json

Memory Object Structure:

Field	Type	Description
memory_text	string	The actual content of the stored memory.
type	string	The category of the memory. Returns either "core" or "general".
memory_id	string	The unique identifier for this specific memory entry.
userId	string	The user ID associated with this memory.

Example Response:

[
  {
    "memory_text": "User is planning to buy a new cycle.",
    "type": "general",
    "memory_id": "mem_89abc...",
    "userId": "user_123"
  }
]

Error Responses

The API will return standard HTTP status codes along with an error payload if the request fails.

• 400 Bad Request: Returned if the JSON is malformed, neither query nor messages are provided, or if the messages array is empty.
• 500 Internal Server Error: Returned if the backend storage mechanism fails to retrieve the memories.

---

Python SDK

You can retrieve memories using the get_memory method in the Go Memory Python client. The SDK enforces strict validation, raising a ValueError if you attempt to send both messages and query simultaneously.

Usage

Example 1: Using a direct query

from gomemory import GoMemoryClient

client = GoMemoryClient(base_url="http://localhost:9000")

memories = client.get_memory(
    user_id="user_123",
    query="I like dogs",
    threshold=0.7
)

for memory in memories:
    print(f"Memory: {memory.memory_text}, Type: {memory.type}")

Example 2: Using conversation messages

from gomemory import GoMemoryClient, Message, Role

client = GoMemoryClient(base_url="http://localhost:9000")

messages_context = [
    Message(role=Role.USER, content="Need to buy a new cycle"),
    Message(role=Role.MODEL, content="That sounds exciting! Do you have an itinerary yet?")
]

memories = client.get_memory(
    user_id="user_123",
    messages=messages_context,
    threshold=0.65
)

for memory in memories:
    print(f"Retrieved: {memory.memory_text}")