curl --request POST \
--url https://api.memmachine.ai/v2/memories/search \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data @- <<EOF
{
"query": "What was the user's last conversation about finance?",
"project_id": "",
"agent_mode": false,
"expand_context": 0,
"filter": "metadata.user_id=123 AND metadata.session_id=abc",
"score_threshold": 0,
"set_metadata": {},
"top_k": 5,
"types": [
"episodic",
"semantic"
]
}
EOF{
"content": {
"episodic_memory": {
"long_term_memory": {
"episodes": [
{
"content": "<string>",
"producer_id": "<string>",
"producer_role": "<string>",
"uid": "<string>",
"created_at": "2023-11-07T05:31:56Z",
"episode_type": "message",
"metadata": {},
"produced_for_id": "<string>",
"score": 123
}
]
},
"short_term_memory": {
"episode_summary": [
"<string>"
],
"episodes": [
{
"content": "<string>",
"producer_id": "<string>",
"producer_role": "<string>",
"uid": "<string>",
"created_at": "2023-11-07T05:31:56Z",
"episode_type": "message",
"metadata": {},
"produced_for_id": "<string>",
"score": 123
}
]
}
},
"semantic_memory": [
{
"category": "<string>",
"feature_name": "<string>",
"tag": "<string>",
"value": "<string>",
"metadata": {
"citations": [
"<string>"
],
"id": "<string>",
"other": {}
},
"set_id": "<string>"
}
]
},
"status": 0
}Search Memories
Search memories within a project.
System returns the top K relevant memories matching the natural language query. The result is sorted by timestamp to help with context.
The filter field allows for filtering based on metadata key-value pairs. The types field allows specifying which memory types to include in the search.
curl --request POST \
--url https://api.memmachine.ai/v2/memories/search \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data @- <<EOF
{
"query": "What was the user's last conversation about finance?",
"project_id": "",
"agent_mode": false,
"expand_context": 0,
"filter": "metadata.user_id=123 AND metadata.session_id=abc",
"score_threshold": 0,
"set_metadata": {},
"top_k": 5,
"types": [
"episodic",
"semantic"
]
}
EOF{
"content": {
"episodic_memory": {
"long_term_memory": {
"episodes": [
{
"content": "<string>",
"producer_id": "<string>",
"producer_role": "<string>",
"uid": "<string>",
"created_at": "2023-11-07T05:31:56Z",
"episode_type": "message",
"metadata": {},
"produced_for_id": "<string>",
"score": 123
}
]
},
"short_term_memory": {
"episode_summary": [
"<string>"
],
"episodes": [
{
"content": "<string>",
"producer_id": "<string>",
"producer_role": "<string>",
"uid": "<string>",
"created_at": "2023-11-07T05:31:56Z",
"episode_type": "message",
"metadata": {},
"produced_for_id": "<string>",
"score": 123
}
]
}
},
"semantic_memory": [
{
"category": "<string>",
"feature_name": "<string>",
"tag": "<string>",
"value": "<string>",
"metadata": {
"citations": [
"<string>"
],
"id": "<string>",
"other": {}
},
"set_id": "<string>"
}
]
},
"status": 0
}Authorizations
Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
Body
Specification model for searching memories.
The natural language query used for semantic memory search. This should be a descriptive string of the information you are looking for.
"What was the user's last conversation about finance?"
The identifier of the project. If empty, the user's default project is used.
255^[\p{L}\p{N}_:-]*$Whether to enable top-level retrieval-agent orchestration for episodic search. When false, episodic search uses direct memory retrieval.
false
The number of additional episodes to include around each matched episode from long term memory for better context.
0
An optional string filter applied to the memory metadata. This uses a simple query language (e.g., 'metadata.user_id=123') for exact matches. Multiple conditions can be combined using AND operators. The metadata fields are prefixed with 'metadata.' to distinguish them from other fields.
"metadata.user_id=123 AND metadata.session_id=abc"
The minimum score for a memory to be included in the search results. Defaults to -inf (no threshold) represented as None. Meaningful only for certain ranking methods.
0
Optional metadata key-value pairs used to filter or identify a specific semantic memory set. Applies only when searching semantic memories.
The maximum number of memories to return in the search results.
5
A list of memory types to include in the search (e.g., episodic, semantic). If empty, all available types are searched.
Memory type.
semantic, episodic ["episodic", "semantic"]