The AI Agent provider exposes several methods that you can call to control the AI Agent's behavior and manage the conversation.
Core methods
| Method | Description | Parameters |
|---|
addUserMessage(message, metadata?) | Adds a user message to the conversation | message: string, metadata?: any |
run() | Starts or continues the AI Agent run | None |
stop() | Stops the current AI Agent run | None |
reset(chatMessages?) | Resets the conversation to its initial state | chatMessages?: ChatMessage[] |
Conversation management
| Method | Description | Parameters |
|---|
setCheckpoint() | Saves the current state as a checkpoint | None |
restoreCheckpoint(checkpoint) | Restores a previously saved checkpoint | checkpoint: Checkpoint |
setChatMessages(messages) | Replaces all chat messages | messages: ChatMessage[] |
setAutoAccept(autoAccept) | Changes the auto-accept setting | autoAccept: "always" | "never" | "onlyRead" |
| Method | Description | Parameters |
|---|
acceptToolCall() | Accepts the pending tool call | None |
rejectToolCall() | Rejects the pending tool call | None |