{"slug": "i-finally-counted-my-tokens-before-they-hatched", "title": "I finally counted my tokens before they hatched", "summary": "Developer Maneshwar built git-lrc, a micro AI code reviewer, and discovered Tokscale, a Rust-based CLI tool by Junhoye that tracks token usage and costs across multiple AI coding agents like Claude Code, Cursor, and Copilot. The tool reads local logs to provide a terminal dashboard with breakdowns by input, output, cache, and reasoning tokens, plus a heatmap of daily AI usage.", "body_md": "*Hello, I'm Maneshwar. I'm building git-lrc, a Micro AI code reviewer that runs on every commit. It is free and source-available on Github. Star git-lrc to help devs discover the project. Do give it a try and share your feedback.*\n\nI basically don't code alone anymore.\n\nCursor is open on my left, Claude is open on my right, and I spend my day describing things to two very confident interns who never sleep.\n\nIt's great.\n\nMy output has genuinely gone up.\n\nThen today evening, mid-prompt, a thought crept in: *how much am I actually spending on all this?* Not the flat subscription.\n\nThe other bill.\n\nThe sneaky usage-based one that quietly climbs while I'm busy telling an AI to \"just refactor this real quick.\"\n\nMy brain did what any healthy developer brain does.\n\nIt panicked, calmed itself with a lie, then panicked harder.\n\nThat little spiral is what sent me looking, and that's how I found\n\nA high-performance CLI tool and visualization dashboard for tracking token usage and costs across multiple AI coding agents.\n\nTip\n\nv2 is here — native Rust TUI, cross-platform support, and more.\n\nI drop new open-source work every week. Don't miss the next one.\n\n[🇺🇸 English](https://github.com/junhoyeo/tokscale/README.md) | [🇰🇷 한국어](https://github.com/junhoyeo/tokscale/README.ko.md) | [🇯🇵 日本語](https://github.com/junhoyeo/tokscale/README.ja.md) | [🇨🇳 简体中文](https://github.com/junhoyeo/tokscale/README.zh-cn.md)\n\nRun[to submit your usage data to the leaderboard and create your public profile!]`bunx tokscale@latest submit`\n\n**Tokscale** helps you monitor and analyze your token consumption from:\n\n| Logo | Client | Data Location | Supported |\n|---|---|---|---|\n|\n\n`~/.local/share/opencode/opencode.db`\n\n(1.2+, all channels including `opencode-stable.db`\n\n) or/and `~/.local/share/opencode/storage/message/`\n\n(legacy/unmigrated)`~/.claude/projects/`\n\nand `~/.claude/transcripts/`\n\n`~/.openclaw/agents/`\n\n(+Short version: it's a CLI that adds up your token usage and cost across pretty much every AI coding tool you already use.\n\nClaude Code, Cursor, Codex, Gemini, Copilot, Amp, and a long list of others I've never even opened.\n\nThe clever bit is that it doesn't ask you to wire up API keys or babysit some dashboard.\n\nAll these tools already write logs to your machine.\n\ntokscale just reads what's sitting there, prices it against live rates, and hands you the damage in a terminal UI.\n\nNo account required to see your own numbers.\n\nHere's the flow, because I think it's cooler than it has any right to be:\n\nThe core is written in Rust, so parsing months of logs is fast enough that you won't go make coffee waiting for it.\n\nIt breaks everything down by input, output, cache reads and writes, even reasoning tokens, so you can finally see which model is quietly eating your lunch.\n\nTrying it is one line. Nothing to install permanently:\n\n```\nnpx tokscale@latest\n```\n\nThat drops you straight into the interactive dashboard.\n\nThere's a GitHub-style contribution graph in there too, except instead of green squares for commits, you get a heatmap of how hard you leaned on the robots each day.\n\nHumbling. Beautiful. Slightly incriminating.\n\nHere's the part that made me smile. tokscale is named after the [Kardashev scale](https://en.wikipedia.org/wiki/Kardashev_scale), the astrophysics idea that ranks a civilization by how much energy it can harness.\n\nType I bends a whole planet's energy. Type II captures an entire star. Type III commands a galaxy.\n\nThe pitch is that in AI-assisted dev, tokens are the new energy. So tokscale ranks *you* on the same idea:\n\nI opened the tool fully expecting to be a humble Type I. A polite, energy-conscious little developer.\n\nReader, I was not.\n\nHonestly, yes. Not because the number was scary (okay, it was a little scary), but because I finally had a real number instead of a vibe.\n\nI could see that most of my spend came from one specific \"just one more prompt\" habit late at night, and that one model was doing the heavy lifting while I paid premium rates for another out of pure muscle memory.\n\nThat's the actual value here. It turns \"eh, probably fine\" into a receipt you can act on.\n\nA token of appreciation, if you will, for your own bank account.\n\nIf you're pair-programming with AI every day like I am, spend the sixty seconds.\n\nRun it, poke around the dashboard, and find out which Kardashev type you really are.\n\nWorst case, you learn something.\n\nBest case, you save enough to justify the subscription that got you here.\n\nYou can grab it on [the repo](https://github.com/junhoyeo/tokscale), and there's a slick web version of the whole thing to explore too.\n\nNow if you'll excuse me, I have a prompt to send. Just one more.\n\nAI agents write code fast. They also silently remove logic, change behavior, and introduce bugs — without telling you. You often find out in production.\n\ngit-lrc fixes this. It hooks into git commit and reviews every diff before it lands. 60-second setup. Completely free.\n\nAny feedback or contributors are welcome! It's online, source-available, and ready for anyone to use.\n\n⭐ Star it on GitHub:\n\n| [🇩🇰 Dansk](https://github.com/HexmosTech/git-lrc/readme/README.da.md) | [🇪🇸 Español](https://github.com/HexmosTech/git-lrc/readme/README.es.md) | [🇮🇷 Farsi](https://github.com/HexmosTech/git-lrc/readme/README.fa.md) | [🇫🇮 Suomi](https://github.com/HexmosTech/git-lrc/readme/README.fi.md) | [🇯🇵 日本語](https://github.com/HexmosTech/git-lrc/readme/README.ja.md) | [🇳🇴 Norsk](https://github.com/HexmosTech/git-lrc/readme/README.nn.md) | [🇵🇹 Português](https://github.com/HexmosTech/git-lrc/readme/README.pt.md) | [🇷🇺 Русский](https://github.com/HexmosTech/git-lrc/readme/README.ru.md) | [🇦🇱 Shqip](https://github.com/HexmosTech/git-lrc/readme/README.sq.md) | [🇨🇳 中文](https://github.com/HexmosTech/git-lrc/readme/README.zh.md) | [🇮🇳 हिन्दी](https://github.com/HexmosTech/git-lrc/readme/README.hi.md) |\n\nGenAI today is a **race car without brakes**. It accelerates fast -- you describe something, and large blocks of code appear instantly. But AI agents *silently break things*: they remove logic, relax constraints, introduce expensive cloud calls, leak credentials, and change behavior -- without telling you. You often find out in production.\n\n** git-lrc is your braking system.** It hooks into\n\n`git commit`\n\nand runs an AI review on every diff In short, git-lrc helps **Prevent Outages, Breaches, and Technical Debt Before They Happen**\n\n**At a glance:** [10 risk categories](https://github.com/HexmosTech/git-lrc#what-git-lrc-checks-for) · [100+ failure patterns tracked](https://github.com/HexmosTech/git-lrc#what-git-lrc-checks-for) · every commit…", "url": "https://wpnews.pro/news/i-finally-counted-my-tokens-before-they-hatched", "canonical_source": "https://dev.to/lovestaco/i-finally-counted-my-tokens-before-they-hatched-32i4", "published_at": "2026-07-11 17:49:54+00:00", "updated_at": "2026-07-11 18:22:08.336695+00:00", "lang": "en", "topics": ["developer-tools", "ai-tools", "artificial-intelligence", "large-language-models"], "entities": ["Maneshwar", "Junhoye", "Tokscale", "Claude Code", "Cursor", "Copilot", "Gemini", "Codex"], "alternates": {"html": "https://wpnews.pro/news/i-finally-counted-my-tokens-before-they-hatched", "markdown": "https://wpnews.pro/news/i-finally-counted-my-tokens-before-they-hatched.md", "text": "https://wpnews.pro/news/i-finally-counted-my-tokens-before-they-hatched.txt", "jsonld": "https://wpnews.pro/news/i-finally-counted-my-tokens-before-they-hatched.jsonld"}}