-
-
Notifications
You must be signed in to change notification settings - Fork 5
Chat History Node
Vitor Lima edited this page Mar 19, 2026
·
1 revision
The Chat History node retrieves messages from the current chat. Use it to feed conversation context into an LLM Agent or JavaScript node.
| Input | Type | Required | Description |
|---|---|---|---|
| Participant ID | String | No | Filter messages to only those from a specific participant. Leave unconnected to include messages from everyone. |
| Output | Description |
|---|---|
| Chat History | A list of chat messages, filtered and limited by your configuration. |
Click the settings icon to configure:
- Node Name — A label to help you identify this node on the canvas.
- Message Depth — How many recent messages to retrieve (1–500). Default is 10. The node always takes the most recent messages.
-
Message Type — Filter by sender:
- All Messages — Everything (default).
- Assistant Only — Only character/AI responses.
- User Only — Only what the user said.
- System Only — Only system messages.
- The node reads all messages from the active chat's current chapter.
- If a Participant ID is connected, it filters to only messages from that participant.
- It applies the Message Type filter from your configuration.
- It takes the last N messages based on the Depth setting.
- The resulting message list flows out.
- The output is a Message List, not a string. Connect it to an LLM Agent's "Chat History" input or a JavaScript node for processing.
- Depth controls token usage. More messages means more context for the AI, but also more tokens. Start with 10–20 and adjust based on your needs.
- Combine with Participant Picker to get only one character's messages — useful for summarization or analysis tasks.