AI Usage Tracker - One local dashboard for Claude and Codex spend

Hey Product Hunt — I've been building this as Claude Usage Tracker. Today it becomes AI Usage Tracker. When I first launched it, almost everyone I built it for was Claude-only. That's not true anymore. People run Codex CLI for one task, Claude Code for the next, Cursor with Sonnet for editing, Codex Exec for batch work — sometimes in the same project, the same day. So the question is no longer "how much did I spend on Claude?" It's "what did my AI tooling cost me — across providers — and where did the money go?" v3.0 answers that: - Codex tracking. New parser for ~/.codex/sessions/ rollout files, priced against OpenAI's standard pricing (GPT-5.5, 5.4, 5.4 Mini, 5.3 Codex, 5.2). Reasoning tokens are shown separately in session detail for visibility but not double-counted — OpenAI already bills them inside output_tokens. - Provider pill. One All / Claude / Codex toggle scopes everything — top-line stats, daily spend chart, project rollup, heatmap, session log. - One combined dollar figure. Costs are normalized to API-equivalent USD on both sides, so totals compare cleanly even if you're a Claude Max subscriber on one side and pay-per-token Codex on the other. - Same local-first stance. No account, no cloud, no telemetry. Source still MIT. $9 signed macOS build for the easiest install. - Auto-migration for 2.x users on first launch. Supported today: Claude Code CLI, Claude Desktop, Cursor, Windsurf, Cline, Roo Code, Aider, Continue.dev, OpenClaw/Clawdbot, and Codex CLI/Exec/Review. Two things I'd love feedback on: 1. If you use both Claude and Codex daily, does one combined dollar figure change how you decide which to reach for? 2. What provider should land next — Gemini? Copilot? Something else? Try it: https://usages.pro · Source: github.com/658jjh/claude-usage-tracker