Reports actual session token usage and savings from the Claude Code session log.
npx skills add https://github.com/TrayMachi/dotfiles --skill caveman-stats-traymachi
Please help me install this Agent Skill. Skill: caveman-stats Source: https://github.com/TrayMachi/dotfiles/tree/main/agents/skills/caveman-stats Command: npx skills add https://github.com/TrayMachi/dotfiles --skill caveman-stats-traymachi
⛏ 12.4k). - Use Case: After a long coding session, run /caveman-stats to see exactly how many tokens were used and how many were saved compared to a standard session. ## Quick Start Type /caveman-stats in your Claude Code session to display the current session's real token usage and savings.High-intent search queries and answers about installing and using this skill.
Run /caveman-stats during the session. The caveman-mode-tracker hook intercepts the command, reads the JSONL session log on disk, and returns formatted input/output token counts plus estimated savings.
The hook parses the session log directly to count actual input and output tokens, then compares output against an estimated non-caveman baseline. The model itself never computes the numbers.
No. The hook returns decision: "block" with the formatted stats as the reason, so the user sees the numbers immediately without the model performing any computation.
Each run writes a lifetime-savings suffix file that the statusline badge reads to display cumulative savings, for example `⛏ 12.4k`.
Model-reported figures are estimates, while caveman-stats reads the actual JSONL session log on disk. The log-based numbers are the authoritative token counts for the session.
rvdbreemenReport measured token usage and savings from Claude Code JSONL session logs.
yeaboi-aiGenerates per-model, per-project AI agent cost reports from local Claude Code session logs.
Ab0umarReports actual session token usage and savings from the Claude Code session log.
se1111thProxy external API calls through Replit-managed billing with query parameters.