caveman-stats

Parse session JSONL logs to report token usage and cost savings.

Updated Jul 6, 2026
One-click install
npx skills add https://github.com/shirulot/codex-skill --skill caveman-stats-shirulot
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: caveman-stats
Source: https://github.com/shirulot/codex-skill/tree/main/caveman-stats
Command: npx skills add https://github.com/shirulot/codex-skill --skill caveman-stats-shirulot

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill solves the lack of transparency regarding token consumption during AI sessions by providing precise, log-based usage metrics instead of relying on potentially inaccurate AI estimations.

Core Features & Use Cases

  • Real-time Token Reporting: Displays exact input and output token counts derived directly from session JSONL logs.
  • Savings Visualization: Calculates and displays the estimated token savings achieved by using the caveman-mode-tracker.
  • Use Case: Use this during a long coding session to monitor your token budget and verify the efficiency gains of your current interaction mode.

Quick Start

Type the command /caveman-stats into the chat interface to display your current session token usage and estimated savings.

Frequently Asked Questions about caveman-stats

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I track real-time token usage during a coding session?

You can track real-time token usage by parsing local session JSONL log files to extract exact input and output token counts. This ensures accurate metrics by avoiding model-based estimation errors.

How do I monitor token consumption without relying on AI estimations?

Monitor token consumption by reading session JSONL log files directly to calculate precise usage metrics. This intercepts user queries and injects performance statistics as a system-level response for data integrity.

Can I visualize token savings within the Claude Code environment?

Yes, you can visualize estimated token savings within the Claude Code environment. It calculates savings achieved by analyzing the session log files and displays the efficiency gains of your current interaction mode.

Does token usage tracking require direct access to session log files?

Yes, token usage tracking requires direct read access to session JSONL log files. This access is necessary to ensure data integrity and provide accurate cost-saving analysis instead of estimations.

What is the best way to verify token budget efficiency during long coding sessions?

The best way to verify token budget efficiency is by parsing session JSONL logs to display exact token counts and estimated savings. This provides precise, log-based usage metrics during long coding sessions.

Why are my AI-based token usage metrics inaccurate?

AI-based token usage metrics are inaccurate because they rely on model estimations. Parsing local session JSONL log files directly ensures data integrity and provides exact input and output token counts.