itchernetski

itchernetski / claude-code-token-meter

Public

Local-first dashboard for Claude Code token usage. Parses ~/.claude/projects/*.jsonl and serves an interactive viewer on localhost. No API calls.

github.comitchernetskiclaude-code-token-meter analytics claude claude-code dashboard fastapi
13
0
100% credibility
Found May 17, 2026 at 14 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A free, local-first dashboard that reads your Claude Code session logs and transforms them into beautiful, interactive analytics. It shows you daily token breakdowns, which projects consume the most resources, how your usage splits across different AI models, and your burn rate over rolling time windows. You can optionally calibrate the dashboard against your Anthropic plan limits to see progress bars. Everything runs entirely on your own computer—no cloud services, no external calls, no data leaves your machine.

How It Works

1
💻 You use Claude Code every day

You rely on Claude Code to help you write and debug code, and you're curious about how much you're actually using.

2
🔍 You want to understand your usage

Claude Code shows you your current burn but doesn't give you a detailed breakdown of your sessions, projects, or spending patterns.

3
📦 You install the tool in seconds

You run a simple install command and the tool is ready to go on your computer.

4
🚀 You launch the dashboard

You run one command and a beautiful dashboard opens in your browser, instantly showing you all your sessions at a glance.

5
📊 You explore your usage patterns

You can see which projects use the most tokens, which days were busiest, and how your usage breaks down between different AI models.

6
You calibrate your plan limits
✋
Manual calibration

You run /usage in Claude Code, copy the percentages, and enter them to set your limits

🔄
Automatic calibration

The tool learns from your rate-limit hits over time and estimates your limits automatically

🎉 You have full visibility into your usage

Everything stays private on your computer, and you can refresh anytime to see updated stats. No cloud, no external calls, just your data.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 14 to 13 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is claude-code-token-meter?

This is a local dashboard that visualizes your Claude Code token consumption by parsing the JSONL log files that Claude Code already writes to disk. It serves a web UI on your localhost that shows daily breakdowns, project-level usage, model distribution (opus/sonnet/haiku), and your top-30 heaviest sessions. The Python backend runs entirely offline, reading from `~/.claude/projects/*/*.jsonl` with zero external API calls. You can calibrate the dashboard against your plan limits using the `/usage` command output, or let it auto-detect your caps from historical rate-limit hits stored in a local SQLite database. The tool exposes a REST API for scripting custom reports and supports both light/dark themes and English/Russian languages.

Why is it gaining traction?

The hook is privacy and zero cost: you see exactly what you are spending without sending data anywhere. Unlike rough estimates or third-party dashboards, this tool reads the authoritative source (your local logs) and applies Anthropic's published cost ratios to give a meaningful relative ranking of sessions and days. The quota calibration feature solves a real pain: Anthropic does not reveal the exact formula for plan limits, so being able to anchor the dashboard to your actual `/usage` percentage and watch progress bars fill up is genuinely useful for budget-conscious developers.

Who should use this?

Claude Code power users on paid plans who want to track their token burn rate day-to-day. Developers working across multiple projects who suspect one repo is chewing through their quota faster than others. Anyone who has hit a rate limit and wanted to understand their consumption patterns retroactively. Teams sharing a single Claude subscription could use this to establish per-project baselines.

Verdict

At 13 stars and 1.0% credibility, this is a small but focused tool with clear documentation and a working feature set. The codebase shows care: FastAPI for the server, SQLite for persistence, clean i18n, and a non-trivial calibration algorithm. It is production-ready for personal use, though the test coverage and long-term maintenance track record are still thin. If you are serious about controlling your Claude Code spend, it is worth installing; just do not bet critical infrastructure on it without reviewing the tests first.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.