Memory Summarization Skill

Capabilities

• Implement conversation summarization strategies
• Configure rolling summary updates
• Design hierarchical summarization
• Implement token-aware summarization
• Create extractive and abstractive summaries
• Design summary quality evaluation

Target Processes

• conversational-memory-system
• long-term-memory-management

Implementation Details

Summarization Strategies

1. Rolling Summary: Update summary with new messages
2. Hierarchical: Multi-level summarization
3. Token-Budget: Fit within token limits
4. Extractive: Key message selection
5. Abstractive: LLM-generated summaries

Configuration Options

• LLM for summarization
• Summary token budget
• Update frequency
• Summary template
• Quality thresholds

Best Practices

• Balance detail vs compression
• Preserve key information
• Monitor summary quality
• Test with long conversations
• Handle context window limits

Dependencies

• langchain-core
• LLM provider