stats-tracker

The Stats Tracker skill provides a comprehensive dashboard and command-line interface for monitoring AI agent activity within the CircleTel development environment. Designed for developers and team leads who utilize Claude Code, this skill transforms raw session data into actionable productivity insights. By tracking metrics such as token usage, model distribution, and daily streaks, it enables teams to make data-driven decisions regarding model selection—choosing between Haiku, Sonnet, and Opus based on cost-efficiency and task complexity. It further enhances developer performance by offering specific reports on 'skill effectiveness' through RSI metrics and pass@k verification logging.

• Real-time monitoring of token consumption, session frequency, and session costs to maintain awareness of usage limits.

• Detailed model usage breakdown, identifying the percentage split between Haiku, Sonnet, and Opus to optimize budget and performance.

• Productivity gamification features, including streak tracking for consistent coding habits and session naming for feature-specific analytics.

• Deep-dive effectiveness dashboards that allow users to drill down into specific skill performance and verify code quality via pass@k metrics.

• Seamless integration with existing tools like ccusage and internal project management workflows for comprehensive end-to-end oversight.

• To use effectively, initiate a session with /stats to check streaks and plan your daily capacity, or use /usage to monitor token limits before starting major architecture tasks.

• Organize complex work by using /rename to assign session names to specific features, enabling granular tracking of usage per feature.

• For performance optimization, default to Sonnet for general coding, Haiku for quick explorations and background tasks, and reserve Opus for complex, high-reasoning architectural problems.

• When encountering high token usage, leverage the stats to identify which sessions are most expensive and adjust your query strategies by being more targeted or splitting sessions.

• Weekly review patterns are encouraged to align token spend with project milestones, using the dashboard as a formal reporting tool for team efficiency.