compact-session

To reduce context window token usage, trigger session-memory compaction to condense message history into summary nodes. This proactively optimizes memory before hitting context limits, preventing performance degradation during long-horizon tasks.

Session memory compaction is a process that condenses message history into summary nodes. It uses the ll-compact-session CLI tool to execute idempotent summarization on the session database, reducing overall context token footprint while maintaining critical context.

You should manually compact session memory when working on long-horizon coding tasks where context window limits are approached. This proactive context management ensures the agent maintains focus without losing critical context as the session grows.

Yes, condensing message history into summary nodes requires access to the session database and the ll-compact-session CLI tool. These dependencies execute the idempotent summarization process needed for effective context optimization.

The best way to free up context space without losing critical history is manual memory compaction. It summarizes the conversation history into summary nodes, allowing you to reduce the token footprint while preserving essential context for the AI agent.

Yes, session compaction supports idempotent summarization through the ll-compact-session CLI tool. This ensures repeated compaction executions on the session database produce consistent summary nodes without duplicating data or corrupting the message history.