yaojingang

yaojingang / tokkit

Public

Lightweight local-first usage ledger for AI coding tools: track tokens, cost, models, terminals, and clients from local logs and proxies.

ai-tools developer-tools llm macos sqlite
10
0
100% credibility
Found Apr 07, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

TokKit is a lightweight local tool that collects and reports token usage and estimated costs from AI coding assistants by scanning their local logs, exports, and proxies.

How It Works

1
📖 Discover TokKit

You learn about TokKit, a simple tracker that watches your AI coding helpers to show exactly how much you're using and spending.

2
🚀 Set it up quickly

You get TokKit running on your Mac in just a few minutes with easy steps.

3
🔗 Link your AI tools

You connect tools like Codex, Warp, or Cursor by pointing it to their activity spots on your computer.

4
🔍 Gather your usage

You run a quick scan to pull in recent sessions from your tools into one clear record.

5
📊 See your first report

You view a neat summary of today's tokens, costs, and breakdowns by tool and time – everything in one place!

6
⏰ Turn on daily updates

You set it to automatically check and save reports every day so you never miss your stats.

🎉 Track like a pro

Now you have honest, private tallies of your AI usage and spending, helping you stay in control effortlessly.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 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 tokkit?

Tokkit is a lightweight Python CLI that aggregates token usage, costs, models, terminals, and clients across AI coding tools like Codex, Claude Code, Warp, Kaku, Cursor, and Copilot by scanning local logs and proxies into a single SQLite ledger. Run `tok today` for daily breakdowns by hour, terminal, model, or source, or `tok last 7` for trend charts and budgets—no cloud dashboards or SDKs needed. It solves the pain of chasing fragmented vendor metrics on one machine, delivering honest reports labeling data as exact, partial, or estimated.

Why is it gaining traction?

Unlike heavy observability stacks requiring app ownership and instrumentation, tokkit stays local-first with zero hosting, fast terminal UX like autosuggest and `tok doctor` diagnostics, and low-friction scans via launchd automation on macOS. Developers dig the proxy for exact Kaku tracking and Augment patching, plus customizable pricing overrides for realistic Est.$ costs. As a github lightweight charts python tool for coding clients, it hooks multi-tool users tired of separate dashboards.

Who should use this?

Mac-based devs juggling AI assistants like Cursor, Warp, and Copilot for daily coding workflows, especially those tracking personal budgets across terminals and IDEs. Ideal for solo engineers auditing token spend without vendor lock-in, or teams evaluating lightweight github alternatives to enterprise logging. Skip if you're cloud-only or need production-scale metrics.

Verdict

Try tokkit if you're on multiple AI tools—its alpha status shines in docs and CLI polish, but 10 stars and 1.0% credibility score mean watch for stability. Solid for personal ledgers; pair with `tok setup` for quick wins.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.