Skip to main content
POST

Overview

Reads a conversation and extracts the durable facts, preferences, decisions and so on as typed memories, using an LLM. With dry_run: true it returns the candidates without storing them. Stored memories get the tag conversation-extract.

Authentication

string
required
Your Memanto Cloud API key (mk_...). See Authentication.

Path Parameters

string
required
The agent’s ID.

Body

array
required
The conversation: objects with role and content.
boolean
default:"false"
Return candidates without storing them.
integer
default:"20"
Most memories to extract.
string
Model to extract with. See List Models.
With dry_run: true the response has candidates and count only. Extracted memories get source: "conversation".