Analyzes Claude Code session bloat and monitors context window usage to prevent performance degradation.
This skill provides deep visibility into the resource consumption of your Claude Code sessions, identifying exactly where tokens and memory are being spent. By breaking down message types, identifying large contributors to session bloat, and providing visual usage metrics, it helps developers maintain high-performance AI interactions. It automates the process of auditing session 'weight' and suggests specific pruning prescriptions—gentle, standard, or aggressive—to keep sessions lean, protect Agent Teams from context loss, and optimize token efficiency.
Key Features
01Automated 'vital signs' check for progress ticks and system reminders.
02Identification of the top 10 largest messages causing session bloat.
03Detailed breakdown of message types including file history, thinking content, and tool results.
04182 GitHub stars
05Intelligent prescription recommendations based on total session byte size.
06Real-time token count tracking and 200K context window percentage visualization.
Use Cases
01Auditing a long-running session to determine why response latency is increasing.
02Determining the appropriate pruning level needed before starting a complex coding task.
03Identifying specific large file snapshots or tool outputs that are consuming excessive context.
