mind.remember
To store information in the mind, use the remember
method. This allows the Mind to persist knowledge and generate a system message that can be passed to the LLM.
const response = await mind.remember([
{
"author": "user",
"message": "I want to learn Python and JavaScript"
},
{
"author": "assistant",
"message": "I understand you want to learn more programming languages"
}
]);
Arguments
messages
Array<{author: string, message: string}>
Array of message objects containing the conversation history to remember. Each message must have an author
and message
field.
Response
metadata
object
Configuration and identifying information about a mind (see metadata structure below).
systemMessage
string
A generated message that provides context and relevant knowledge for the mind's responses.
Metadata Structure
relatedKnowledge
string
Contains any related knowledge found in the mind's memory.
rawMessage
string
The original message in JSON format containing author and message.
messageWithContext
string[]
Array of messages with additional context added.
updateValues
string[]
Paths in the mind's memory structure where this information was stored (e.g., "/unstructured", "structured/Coding Languages/unstructured").
changes
object
Object containing any changes made to the mind's memory structure.
// Example metadata response
{
"metadata": {
"relatedKnowledge": "",
"rawMessage": "{\"author\":\"assistant\",\"message\":\"A function is a block of code designed to perform a particular task. A class is a blueprint for creating objects.\"}",
"messageWithContext": [
"A function is a block of code designed to perform a particular task, while a class is a blueprint for creating objects."
],
"updateValues": [
"/unstructured",
"structured/Coding Languages/unstructured"
],
"changes": {}
}
}
Last updated