Workshop on Context Engineering Strategies for Long-Term Model Memory
August 17, 2026
This workshop covers techniques for managing finite context windows and stateless models through compaction methods like truncation, summarization, and tool-result clearing. It demonstrates implementing persistent cross-session memory and on-demand skill loading using an open-source AI tutor.
HOW THIS AFFECTS YOU
●
builderYou can apply these compaction toolkits to reduce token costs and prevent model forgetting in production sessions.
●
researcherThis provides a framework for studying the trade-offs between different context management strategies.