Memory Rotate & Summarize
Automates the summarization and archival of AI session memory files to maintain clean and efficient context windows.
Memory Rotate is a specialized Claude Code skill designed to optimize the way AI agents handle long-term project context. When a memory file is rotated out of the active context, this skill automatically invokes a lightweight Haiku subagent to generate a concise summary of the archived session. By converting raw logs into structured JSON summaries and updating a centralized index, it ensures that critical project history remains searchable and retrievable without consuming unnecessary token space in future sessions.
主要功能
01Centralized indexing for easy context retrieval
02Lightweight Haiku-powered session summarization
03Automated failure recovery and retry logic
04Automatic trigger-based memory log processing
05Structured JSON archival for machine-readable history
060 GitHub stars
使用场景
01Archiving project milestones and technical decisions for future reference
02Managing long-running development sessions without losing key historical context
03Optimizing token usage by summarizing historical logs instead of loading full transcripts