Skip to main content Configure whether the agent maintains conversation history across interactions and how to manage long conversations.
Configuration Enable Conversation History - Toggle switch to enable or disable conversation history.When enabled, the agent will:
Store conversation messages
Use previous context in subsequent interactions
Maintain state across multiple conversation turns
When disabled, each interaction is treated as a new conversation without context from previous messages.
Summarization When conversation history is enabled, you can configure summarization to manage long conversations and reduce token usage.
LLM-Based Summarization Uses an LLM to intelligently summarize older conversation history while preserving important context.
Configuration:
Summarizer Model - LLM provider and model for summarizationSummarizer Prompt - Instructions for how to summarize conversationsToken Threshold - Summarization triggers when total tokens exceed this valueKeep Recent Count - Number of recent conversation runs to keep unsummarized
How it works:
Monitors total token count in conversation history
When threshold is exceeded, keeps the most recent N runs intact
Summarizes older runs into a single system message
Preserves important context while reducing token usage
Sliding Window Summarization Keeps only the most recent N conversation runs and discards older ones. This is a simple, cost-effective approach that doesn’t require an LLM.
Configuration:
Keep Count - Number of recent conversation runs to retain
How it works:
Messages are grouped by conversation run ID
Only the most recent N runs are kept
Older runs are discarded without summarization