skills.rest
ebarroso12
ebarroso12COMMUNITY
Explore all skills by this author →
GitHub

caveman-stats

Report real input and output token usage from Claude Code session JSONL logs.

Updated Aug 23, 2026
Search Tags:#productivity#Claude Code#token usage#cost savings#session logs#prompting
One-click install
npx skills add https://github.com/ebarroso12/personal-skills --skill caveman-stats-ebarroso12
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: caveman-stats
Source: https://github.com/ebarroso12/personal-skills/tree/main/caveman-stats
Command: npx skills add https://github.com/ebarroso12/personal-skills --skill caveman-stats-ebarroso12

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

It removes guesswork from “caveman” usage by showing the actual input/output token counts and the resulting savings for your current session rather than relying on estimates.

Core Features & Use Cases

  • Real token receipts: Reports input and output tokens taken directly from the on-disk Claude Code session JSONL log (no model-based estimation).
  • Savings vs baseline: Computes savings versus a non-caveman baseline using the recorded session data.
  • Contextual injection: Activates when you run /caveman-stats, and the hook returns blocked-decision output containing the formatted stats and a lifetime-savings suffix used by a statusline badge.

Quick Start

Run the command /caveman-stats during your session to immediately display real 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 token usage and cost savings in a Claude Code session?▼

Token usage tracking in Claude Code is done by reading the local on-disk JSONL session log to report exact input and output token counts, comparing them against a non-caveman baseline to calculate cost savings without model estimation.

What is the best way to calculate token savings versus a non-caveman baseline?▼

Token savings versus a non-caveman baseline are calculated by extracting actual input and output token counts from the on-disk JSONL session log and comparing them to the baseline data recorded for the current session.

Does Claude Code session logging store enough data to show exact token receipts?▼

Claude Code session logging does provide exact token receipts by recording actual input and output tokens in the on-disk JSONL log, which can be read directly to report usage without relying on model-based estimation.

Can I monitor token-limited prompting performance without doing token estimation in the model?▼

Monitoring token-limited prompting performance without model estimation is possible by reading the on-disk JSONL session log to retrieve real input and output token counts and compute savings against a baseline.

Why does my prompt token estimation not match the actual Claude Code session logs?▼

Prompt token estimation often differs from actual Claude Code session logs because estimations are model-based, while session logs record the true input and output token counts directly from the on-disk JSONL file.

Related Skills

Explore More Data & Analytics Skills→
junechen7414junechen7414

caveman-stats

Report actual Claude Code token usage from the JSONL session log.

Community
Intermediate
dwaynimaydwaynimay

caveman-stats

Report actual input and output tokens from the Claude Code session JSONL log.

Community
Basic
selfagencyselfagency

caveman-stats

Report actual Claude Code session token usage and savings from JSONL logs.

Community
Basic
jchildreejchildree
1

caveman-stats

Monitor token usage and savings in Claude Code sessions from logs.

Community
Basic