Create conversation
editCreate conversation
editCreate a new Elastic AI Assistant conversation.
Request URL
editPOST <kibana host>:<port>/api/security_ai_assistant/current_user/conversations
Request body
edit| Name | Type | Description | Required |
|---|---|---|---|
|
String |
Conversation title. If you set it to "New chat", the AI will generate a title. |
Yes |
|
String |
Can be "assistant", "insights", or not defined. |
No |
|
Boolean |
Define if conversation is a system conversation which cannot be deleted. Defaults to false. |
No |
|
Boolean |
Defines if conversation can appear as the latest conversation. |
No |
|
Conversation configuration. |
No |
|
|
Array of conversation messages. |
No |
|
|
Key, Value(String, String) |
List of the fields with anonymization. |
No |
apiConfig object
edit| Name | Type | Description | Required |
|---|---|---|---|
|
String |
Kibana connector ID. |
Yes |
|
String |
Kibana connector action type ID. |
Yes |
|
String |
Default system prompt ID. |
Yes |
|
String |
Specific LLM name. |
No |
messages object
edit| Name | Type | Description | Required |
|---|---|---|---|
|
String |
Message role. Can be "user", "assistant", or "system". |
Yes |
|
String |
Message content to send to LLM. |
Yes |
|
Boolean |
Define if the message is an error message instead of an LLM response. |
No |
|
String |
Timestamp when the message was sent. |
No |
Example requests
editExample 1
Creates a new conversation.
POST api/security_ai_assistant/current_user/conversations
{
"title": "The conversation title.",
"category": "assistant",
"messages": [
{
"content": "test content",
"role": "user",
"isError": false,
"timestamp": "2019-12-13T16:40:33.400Z",
"traceData": {
"traceId": "1234",
"transactionId": "2"
}
}
],
"apiConfig": {
"actionTypeId": ".gen-ai",
"connectorId": "86ab-471c-a00b-25b7e20c2d12",
"defaultSystemPromptId": "Default",
"model": "gpt-4o"
},
"isDefault": false,
"excludeFromLastConversationStorage": true,
"replacements": {
"field1": "914beb92-86ab-471c-a00b"
}
}
Response code
edit200
Indicates a successful call.
Response payload
editA JSON conversation object with a unique id.
Example 1
Conversation response payload:
{
"id": "07805df2-6462-451a-b534-78da47873c42",
"title": "The conversation title.",
"category": "assistant",
"timestamp": "2024-07-29T06:58:15.670Z",
"updatedAt": "2024-07-29T06:58:15.670Z",
"createdAt": "2024-07-29T06:58:15.670Z",
"replacements": {
"field1": "914beb92-86ab-471c-a00b"
},
"users": [
{
"name": "elastic"
}
],
"messages": [
{
"content": "test content",
"role": "user",
"timestamp": "2019-12-13T16:40:33.400Z",
"traceData": {
"transactionId": "2",
"traceId": "1234"
}
}
],
"apiConfig": {
"connectorId": "86ab-471c-a00b-25b7e20c2d12",
"actionTypeId": ".gen-ai",
"defaultSystemPromptId": "Default",
"model": "gpt-4o"
},
"isDefault": false,
"excludeFromLastConversationStorage": true,
"namespace": "default"
}