session-management

Manage AI context window growth by monitoring token usage thresholds and proactively pruning irrelevant session history. This Skill provides tools to summarize and optimize context, maintaining performance and controlling costs during complex, long-running tasks.

Recall previous AI session details for multi-day tasks using session navigation tools. You can discover past sessions with `list_sessions`, read specific histories with `read_session`, and search older discussions with `search_session` to quickly grasp previous implementation details.

Start pruning context to optimize AI agent performance when token usage reaches critical thresholds like 70%, 85%, or 95%. Employing summarization strategies at these points helps maintain efficient context and prevents escalating costs.

Yes, this session management approach is specifically applicable to long-running or complex workflows where maintaining performance is critical. It satisfies requirements for efficient context transfer and proactive context pruning strategies.

The best way to summarize older AI discussions is by using the built-in `summarize_session` tool. This allows you to quickly grasp the gist of previous sessions and maintain an efficient context before continuing your work.

A limitation of proactive context pruning strategies is the risk of removing relevant information needed for future tasks. However, employing session summarization alongside pruning mitigates this by preserving the core context and implementation details.