Generates comprehensive token consumption reports and context health analysis for proactive resource management.
The Token Usage Reporter provides deep visibility into the resource consumption of Claude Code subagents by analyzing session files for token usage, duration, and tool invocation counts. It acts as a diagnostic layer that detects when context limits are approaching or when inefficient patterns—such as excessive input-to-output ratios or context compaction—occur. By providing real-time health status indicators and data-driven recommendations for task decomposition, this skill ensures that complex AI orchestrations remain performant and stay within model window constraints.
Key Features
01Proactive warning system based on 40% and 75% thresholds
02Actionable recommendations for task decomposition and efficiency
03Subagent-level token and duration tracking
04Detection and flagging of context compaction events
052 GitHub stars
06Visual context health status indicators (OK, HIGH, EXCEEDED)
Use Cases
01Analyzing subagent efficiency after completing complex multi-step tasks
02Performing periodic health checks during long-running AI orchestrations
03Identifying when to decompose a large task into smaller, focused subagents
