explaingit

keros68/metrik

Analysis updated 2026-05-18

★ 18RustAudience · developerComplexity · 2/5LicenseSetup · moderate

TLDR

A local-first desktop app that tracks your AI coding agent usage (Claude, Codex, and more) by reading logs already on your computer, keeping quota, token counts, and cost estimates separate.

Mindmap

mindmap
  root((Metrik))
    What it does
      Tracks AI agent usage
      Reads local log files
      Separates quota token cost
    Tech stack
      Rust
      Tauri
      TypeScript
    Platforms
      Windows widget
      macOS menu bar
    Supported agents
      Claude
      Codex
      OpenCode
      Kimi
    Use cases
      Monitor quota usage
      Estimate costs
      Sync across devices
    Privacy
      Metadata only
      No prompts stored
      Local first

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Track how much of your Claude Code or Codex usage quota you have left without leaving your desktop.

USE CASE 2

Compare estimated API costs across multiple AI coding agents you use day to day.

USE CASE 3

Sync usage stats across your laptop and desktop through a shared cloud folder you already use.

USE CASE 4

Export a CSV of daily coding-agent sessions to review or share with a team.

What is it built with?

RustTauriTypeScript

How does it compare?

keros68/metrikadindazu/ultimatevocalapplicative-systems/gcan
Stars181818
LanguageRustRustRust
Setup difficultymoderatemoderateeasy
Complexity2/52/52/5
Audiencedeveloperdeveloperops devops

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Installers are unsigned, so you must manually allow the app to run on first launch.

Free to use, modify, and run for yourself. If you distribute a modified version or offer it as a network service, you must also release your changes under the same AGPL license.

In plain English

Metrik is a desktop app that tracks how much you are using AI coding agents like Claude Code or Codex, and it keeps everything on your own machine instead of sending data to a server. It reads the log files that these agents already leave behind on your computer, then shows you three separate kinds of numbers: the official usage quota reported by the agent, the token counts it parsed from your local logs, and an estimated cost based on public pricing. The app is careful to keep these three numbers apart, since the estimated cost is a guess, not a real bill. On Windows, Metrik shows up as a small 320 by 320 pixel widget that can also collapse into a thin horizontal or vertical strip showing quota for each agent as a colored bar. On macOS it lives in the menu bar instead. Either version can expand into a full statistics view with a 26 week heatmap, a weekly trend line, and a breakdown of usage by agent. There is also a daily usage log you can filter, export to CSV, or copy session IDs from so you can resume a session elsewhere. Metrik currently supports ChatGPT and Codex, Claude, ZCode and GLM, OpenCode, Kimi, and Antigravity, though not Gemini CLI or Cursor yet. For syncing across devices, it writes to a shared folder you already use, such as Nutstore, OneDrive, or Syncthing, and merges the exported statistics from each device automatically rather than running its own cloud service. On privacy, the app says it only stores metadata such as timestamps, agent name, model, and session IDs, never your prompts, replies, tool output, or credentials. Getting Claude's quota can work two ways: a default method that reads a status line hook with no network requests, or an optional direct login connection that the app warns may run against Anthropic's current subscription terms of service. Building Metrik from source requires Node.js 22 or newer and Rust 1.88 or newer. It is licensed under AGPL-3.0-or-later, meaning any modified or hosted version must also be shared as open source, though versions before 0.10.0 remain under the more permissive MIT license. Installers are unsigned, so you have to manually allow them to run the first time, and only Windows and macOS builds are provided.

Copy-paste prompts

Prompt 1
Help me set up Metrik's statusLine hook so it can read my Claude Code quota locally.
Prompt 2
Explain the difference between Metrik's statusLine and OAuth quota methods and which is safer to use.
Prompt 3
Walk me through building Metrik from source on macOS with Rust and Node.js installed.
Prompt 4
Show me how to configure Metrik to sync usage data between two computers using Syncthing.

Frequently asked questions

What is metrik?

A local-first desktop app that tracks your AI coding agent usage (Claude, Codex, and more) by reading logs already on your computer, keeping quota, token counts, and cost estimates separate.

What language is metrik written in?

Mainly Rust. The stack also includes Rust, Tauri, TypeScript.

What license does metrik use?

Free to use, modify, and run for yourself. If you distribute a modified version or offer it as a network service, you must also release your changes under the same AGPL license.

How hard is metrik to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is metrik for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.

Alternatives & similar repos