Local-first · Privacy-respecting

How much are your AI coding tools actually costing you?

boardfella tracks token consumption and costs across 20+ AI tools. Your data stays on your computer until you connect it. A connected computer sends its daily totals privately to My usage. Only if you choose to publish are aggregated token totals sent to this leaderboard, never prompts, answers, source code or file paths.

Terminal
# Connect this computer
npx boardfella@1.5.22 connect
localhost:3847
12,847,392
Total Tokens
4.2MInput
3.1MOutput
5.5MCache
$47.20Total Cost
156Sessions
23Active Days

Supports 20+ AI tools

Claude Code
Codex
OpenClaw
OpenCode
Hermes
Qoder
Cursor
Copilot
KiloCode
Kelivo
Gemini CLI
Kimi Code
CodeBuddy
Kiro
Grok Build
Antigravity
Roo Code
Zed
Goose
oh-my-pi
pi
Craft
Droid
ZCode
CodeFuse

All the data you need, in one dashboard

Stop switching between tools. boardfella aggregates usage data from all your AI coding tools in one place.

◇

Token Usage Tracking

Daily bar charts showing input, output, cache read/write, and thinking token consumption. Precise to every API call.

$

Cost Estimation & Analysis

Automatic cost calculation from configurable pricing. Analyze spend by tool, model, and project. No billing surprises.

◆

Model Usage Ranking

Which models are called most? How many tokens each? Clear ranking tables help optimize model selection.

⚡

Tool Call Analytics

What operations does AI perform most? Bash, Read, Edit frequency rankings reveal work patterns.

◎

Project-Level Stats

Usage and cost grouped by repository. Which project consumes the most? Data-driven allocation.

▣

Quota Monitoring

Real-time rate limit quota monitoring for Claude Code, Codex, Copilot, and more. View utilization and reset countdowns.

⇅

Multi-Device Sync

Sync across devices via GitHub or S3-compatible storage. Office, home, travel — data always consistent.

↗

Data Export

Export usage data as CSV, JSON, or NDJSON for integration with your existing data pipelines and reporting.

How It Works

01

Run one command

Open a terminal and run npx boardfella@1.5.22 connect. No global install needed. Your browser opens the helper.

02

See your own numbers

The helper reads the logs of your AI tools on this computer and shows your numbers first. Nothing leaves your computer.

03

Publish if you want

Connect your account, check the preview, then release your numbers.

●
Your Data Stays on Your Machine Until You Connect

All parsing and storage happens locally. The SQLite database lives on your disk. A connected computer sends its daily totals privately to My usage, where only you see them. Only the totals you choose to publish go to the leaderboard.

●
No Account Needed Locally

The local dashboard needs no account, no API keys and no cloud service. You only need an account to publish to the leaderboard or to see your usage on the website.

●
Optional Multi-Device Sync

Sync via your own GitHub repo or S3 storage — data always under your control.

Quick Start

One command. Up and running in five minutes.

01
Run one command

Open a terminal and run npx boardfella@1.5.22 connect. Your browser opens the helper.

02
See your own numbers

The helper finds the logs of your AI tools on this computer and shows your numbers first.

03
Publish if you want

Connect your account, check the preview, then release your numbers.

Read the full docs →
Terminal
$npx boardfella@1.5.22 connect
Your browser opens the helper at http://127.0.0.1:3847/connect once it has read your logs (the first time this can take a few minutes)

Run boardfella in a container. Build the image from the repository's Dockerfile; it includes the CLI and web dashboard with volume-mounted data persistence.

Terminal
# Build in the repository root and run (dashboard password required)
docker build -t boardfella .
mkdir -p ~/.boardfella
# Runs as node (uid 1000); on Linux with another uid add --user "$(id -u):$(id -g)"
docker run -d \
  -p 3847:3847 \
  -e BOARDFELLA_DASHBOARD_PASSWORD=change-me \
  -v ~/.boardfella:/root/.boardfella \
  boardfella

boardfella is open source. Contributions of any kind are welcome — submit issues, open PRs, improve docs, or share your experience.

Start tracking your AI coding spend

Open source, free, and local-first. Up and running in five minutes.