🧾 AI Usage Tracker

Track and visualize Claude + Codex usage costs across all your local development tools

Quick Start · Screenshots · Features · Supported Tools · Pricing · Contributing

---

✨ Overview

AI Usage Tracker (formerly Claude Usage Tracker) is a local-first tool that automatically discovers and aggregates your AI coding usage across 10+ development tools — covering both Anthropic Claude and OpenAI Codex. It scans known data directories, parses JSONL/log files, calculates costs using model-specific pricing, and presents everything in a beautiful dark-themed interactive dashboard powered by Chart.js. A top-level All / Claude / Codex pill toggle scopes the entire dashboard to a single provider.

[!TIP] No cloud. No telemetry. No accounts. Everything stays on your machine — your data never leaves your laptop.

This project is open source under the MIT license. You can build and use it yourself for free. If you want the convenience of a ready-to-use signed macOS app with premium licensing, automatic update checks, and one-click updates, you can buy the Premium Build for $9:

Buy Premium Build →

### 🎯 Built for developers - **Auto-discovery** — detects Claude + Codex tools - **Privacy-first** — 100% local, zero telemetry - **Beautiful UI** — dark mode dashboard with charts - **Zero config** — just run and open

### 📦 Works with `OpenClaw` · `Clawdbot` · `Claude Code CLI` · `Claude Desktop` · `Cursor` · `Windsurf` · `Cline` · `Roo Code` · `Aider` · `Continue.dev` · `Codex CLI` · `Codex Exec` · `Codex Review`

---

📸 Screenshots

Dashboard Overview

Top-line stats (Today / Week / Month / All-time / Sessions), provider pill toggle (All / Claude / Codex), daily spend chart with source breakdown, and donut charts for cost-by-source and cost-by-model.

Projects View

Cost grouped by working directory — see which projects burn through your token budget. Filter by source, model, date range, and minimum cost.

Session Log — Timeline

Day-by-day session timeline with expandable rows. Color-coded model chips (Claude + GPT/Codex), full token breakdown (input / output / cache read / cache write), and per-day totals.

Peak Hours Heatmap

Hour × day activity grid revealing your most productive — and most expensive — coding hours.

Session Detail

Drill into any session: token breakdown (including Codex reasoning tokens), conversation preview, and a one-click resume command (claude --resume <id> for Claude Code, codex resume <id> for Codex).

---

🚀 Features

#### 🔍 Discovery - Claude + Codex auto-detection - 10+ supported tools - Silent fallback for missing tools - Smart provider-aware deduplication	#### 📊 Analytics - Daily / weekly / monthly / all-time - Per-provider, per-model, per-source breakdown - Per-project cost rollup - Monthly cost projections - Yesterday delta comparison	#### 🎨 Visualization - Dark-themed dashboard - Chart.js animated charts - Two heatmap views - Animated stat counters - Responsive layouts
#### 🧮 Cost intelligence - Per-million-token pricing - Opus / Sonnet / Haiku tiers (Anthropic USD) - GPT-5.x / Codex tiers (OpenAI API USD) - Cache read / write + reasoning tokens - Most-expensive-session callout	#### 🔎 Filtering & search - Provider pill (All / Claude / Codex) - Multi-criteria filters with chips - Source / model / date range - Minimum cost threshold	#### ⚡ Productivity - Standalone `.app` bundle - Premium build with automatic update checks - Keyboard shortcuts (`Shift+E`) - One-click session resume (Claude + Codex) - Browser-mode fallback

---

🚀 Quick Start

Option 1 — Premium Build (Convenience)

The source code is free, but the premium build is the easiest way to use the app on macOS.

• Price: $9 one-time purchase
• Includes: signed macOS app, license management, automatic update checks, and one-click updates
• Requirements: Node.js v16+ · macOS 12.0+

Buy Premium Build →

After purchase, download the .dmg, open it, drag AI Usage Tracker to Applications, and launch.

Option 2 — Build From Source (Free)

git clone https://github.com/658jjh/claude-usage-tracker.git
cd claude-usage-tracker
./build-app.sh

Then double-click AI Usage Tracker.app — it collects fresh data and renders everything in a native window.

The source build is open source and does not include the premium license gate or premium update checker.

Option 3 — Browser Mode (any OS)

cd src
node collect-usage.js
python3 -m http.server 8765
open http://localhost:8765/dashboard.html

[!NOTE] Upgrading from 2.x? Your data carries over automatically — the app migrates ~/Library/Application Support/ClaudeUsageTracker → AIUsageTracker on first launch.

---

📦 Supported Tools

Tool	Provider	Data Location	Format
OpenClaw / Clawdbot	Claude	~/.openclaw/agents/main/sessions/	JSONL
Claude Code CLI	Claude	~/.claude/projects/	JSONL
Claude Desktop	Claude	~/Library/Application Support/Claude/	JSONL
Cursor	Claude	~/.cursor/projects/	JSONL
Windsurf	Claude	~/.windsurf/	JSONL
Cline	Claude	~/.cline/	JSONL
Roo Code	Claude	~/.roo-code/	JSONL
Aider	Claude	~/.aider/	JSONL (litellm)
Continue.dev	Claude	~/.continue/sessions/	JSON
Codex CLI / Exec / Review	Codex	~/.codex/sessions/	JSONL (rollout-*)

[!NOTE] Tool detection is automatic. If a tool isn't installed or has no data, it's silently skipped.

---

💰 Pricing Models

Costs are calculated using each provider's per-million-token pricing — Anthropic's published USD for Claude, OpenAI API standard USD for Codex. Totals stay comparable in a single dollar figure regardless of which provider you're viewing.

Anthropic Claude

Model	Input	Output	Cache Write	Cache Read
🔴 Opus 5.0	$20.00	$100.00	$25.00	$2.00
🟠 Opus 4.5 — 4.9	$5.00	$25.00	$6.25	$0.50
🟡 Opus 4.0 / 4.1	$15.00	$75.00	$18.75	$1.50
🟢 Sonnet 3.5 — 4.6	$3.00	$15.00	$3.75	$0.30
🔵 Haiku 4.0 / 4.5	$1.00	$5.00	$1.25	$0.10
🟣 Haiku 3.0 / 3.5	$0.25	$1.25	$0.30	$0.03

OpenAI Codex

Model	Input	Output	Cache Read
🟢 GPT-5.5	$5.00	$30.00	$0.50
🟢 GPT-5.4	$2.50	$15.00	$0.25
🟢 GPT-5.4 Mini	$0.75	$4.50	$0.075
🟢 GPT-5.3 Codex	$1.75	$14.00	$0.175
🟢 GPT-5.2	$2.00	$10.00	$0.20

_{All prices in USD per million tokens. OpenAI bills reasoning tokens as part of output_tokens, so they aren't double-counted — the dashboard shows reasoning separately in the session detail modal for visibility only.}

---

🤝 Contributing

Contributions are welcome — bug fixes, new tool integrations, and design improvements all encouraged.

1. Fork the repository
2. Create a feature branch:  git checkout -b feat/my-feature
3. Commit your changes:      git commit -m "feat: add my feature"
4. Push to your fork:        git push origin feat/my-feature
5. Open a Pull Request

Please follow the existing code style and commit message conventions (feat:, fix:, docs:, chore:).

💡 Ideas for contributions

• 🔌 Add support for additional AI tools (Gemini CLI, Copilot CLI, etc.)
• 📱 Improve mobile responsiveness
• 📤 Add data export (CSV, JSON)
• 🔔 Add cost alerts / budget thresholds
• 🐧 Linux / Windows path support
• 🖥️ Electron or Tauri desktop app

---

📄 License

This project is licensed under the MIT License.

---

Built with ❤️ for the AI developer community

☕ Support the project

If this tool saves you time, consider buying the premium convenience build or buying me a coffee:

Buy Premium Build — $9

_{Made by developers, for developers. Star ⭐ the repo if you find it useful.}