Overview
This node integrates with the Zep v3 memory service to provide persistent memory storage for AI agents using conversation threads. It allows retrieving stored conversational context either as raw thread data or as a natural language summary of the thread history. This is useful in scenarios where an AI agent needs to maintain state or recall past interactions across multiple executions, such as chatbots, virtual assistants, or any system requiring contextual awareness over time.
Practical examples include:
- A chatbot that remembers previous user inputs and references them in follow-up responses.
- An AI assistant summarizing prior conversations to provide concise context before continuing interaction.
- Systems that log conversation threads for audit or review purposes while enabling easy retrieval.
Properties
| Name | Meaning |
|---|---|
| Thread ID | Unique identifier for the memory thread, typically the execution ID. |
| Context Retrieval Mode | How the conversation history is retrieved: "Basic" returns raw thread context; "Summary" returns a natural language summary of the thread history. |
| Memory Key | The key under which the memory will be stored in the output JSON. |
Output
The node outputs a single field named according to the configured Memory Key (default: chat_history). This field contains the retrieved memory data from the Zep service.
- If Context Retrieval Mode is set to "Basic", the output contains the raw thread context data.
- If set to "Summary", the output contains a natural language summary of the thread's conversation history.
No binary data output is produced by this node.
Dependencies
- Requires an API key credential for authenticating with the Zep Cloud service.
- The node depends on the external Zep v3 memory API to fetch and summarize conversation threads.
- Proper configuration of the API key credential within n8n is necessary for operation.
Troubleshooting
- Missing Credentials Error: If the node throws an error about missing Zep Cloud credentials, ensure that a valid API key credential is configured and assigned to the node.
- Initialization Failure: Errors during memory initialization may indicate network issues, invalid thread IDs, or API key problems. Verify connectivity and credential validity.
- Unknown Errors: Generic errors during memory creation suggest unexpected issues; reviewing logs and verifying all input parameters can help resolve these.