Repolode
Submit
kelviq

kelviq/tare

A Claude Code skill that reads your local session logs to explain where token usage went and why you hit limits.

View on GitHubWebsite
Stars
296
Forks
13
Open issues
8
Last commit
2 months ago

What is it?

tare is a skill for Claude Code that analyzes the session logs Claude Code already keeps on your machine. You install it once and then ask questions in plain English, such as why you hit a usage limit, which project or model consumed the most tokens, or how full your current 5-hour window is.

The tool deduplicates repeated log entries so totals are not inflated, and it attributes context costs to the tools that caused them, such as large file reads that get re-sent with every later message. It also tracks the rolling usage window and flags patterns like many short parallel sessions that suggest an automated script rather than manual work.

All processing runs locally. The project states that its scripts make no network connections, and shareable summaries contain only totals, dates and tool names, without prompts, file contents or identifiers. It runs on macOS or Linux with Python 3.9+ and currently reads only Claude Code logs.

Key features

  • 01Answers plain-English questions about usage limits and token spending
  • 02Deduplicates Claude Code log entries so token totals are accurate
  • 03Attributes context re-send costs to the tools and files that caused them
  • 04Shows how full the rolling 5-hour usage window is
  • 05Generates an HTML report and a redacted summary safe to share
  • 06Runs entirely locally with no network connections

Use cases

  • Finding out why you were locked out of Claude Code earlier than expected
  • Identifying which project, model or file is consuming most of your quota
  • Checking whether a background script or automation is running Claude Code
  • Comparing usage before and after changing your workflow, such as using /clear

Quick start

  1. 1

    Install the skill in any terminal

    npx skills add kelviq/tare -g -y --copy --agent claude-code
  2. 2

    Start a new Claude Code session and check that tare appears when you type /

  3. 3

    Ask Claude Code a question in plain English, or run the full diagnosis

    /tare

Things to know

  • Works only on macOS or Linux; Windows paths are not supported.
  • Reads only Claude Code's local logs, not other coding agents.
  • On some machines the installer can miss the skill; a troubleshooting step may be needed.

Topics

#claude-code#token-usage#usage-analysis#llm#developer-tools#privacy#cli

FAQ

What is tare?
A Claude Code skill that reads your local session logs to explain where token usage went and why you hit limits. tare is a skill for Claude Code that analyzes the session logs Claude Code already keeps on your machine. You install it once and then ask questions in plain English, such as why you hit a usage limit, which project or model consumed the most tokens, or how full your current 5-hour window is.
Is tare free to use? What license does it have?
tare is open source under the MIT license. Check the license terms before using it in commercial products.
Is tare still maintained?
Updated within 6 months. The last commit to tare was on Aug 28, 2026.
What can I use tare for?
Finding out why you were locked out of Claude Code earlier than expected Identifying which project, model or file is consuming most of your quota Checking whether a background script or automation is running Claude Code Comparing usage before and after changing your workflow, such as using /clear
How do I install tare?
Install the skill in any terminal: npx skills add kelviq/tare -g -y --copy --agent claude-code Start a new Claude Code session and check that tare appears when you type / Ask Claude Code a question in plain English, or run the full diagnosis: /tare
Where can I get tare?
The source code and installation instructions are on GitHub: https://github.com/kelviq/tare

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects