API Reference

Recall Context

Get a prompt-ready context block for an LLM call, rather than a list of records.

Endpoint and request fields

POST/v1/memory/recall

200 OK

Field	Type	Contract
`tenant_id`	string	Required tenant identifier.
`user_id`	string	Required end-user identifier.
`prompt`	string	Required prompt the context is being assembled for.
`k`	integer	Optional ceiling on memories considered.
`types`	string[]	Optional content-type filter.

Build context

import os
from memorysync import MemorySyncClient

client = MemorySyncClient(
    api_key=os.environ["MEMORYSYNC_API_KEY"],
    base_url="https://api.memorysync.io",
    project_id=os.environ["MEMORYSYNC_PROJECT_ID"],
    end_user_id="usr_7f3a9c2e",
)

result = client.recall(
    tenant_id="acme",
    user_id="usr_7f3a9c2e",
    prompt="Book me a flight to Berlin",
    types=["preference", "fact"],
)
prompt = f"{result['context']}\n\nUser: Book me a flight to Berlin"

Result handling

Returns a context string ready to place in a prompt, plus the sub-queries, intents, entities and cluster hits that produced it. An empty context is a valid result for a user with nothing relevant stored.

How the types filter behaves

Errors and next action

A 422 means a missing required field. For 429 or 5xx, call the model without memory context rather than failing the user's request.

Safety notes

Compose Prompt

Fill a template instead of a raw block.

Query Memory

Get records rather than a block.

Organization Ontology

Make a custom type filterable.

← Previous

Add Conversation Turn

Memory Status