tokenmap

Your AI coding stats, visualized.

A GitHub-style contribution heatmap that shows how much you actually use AI coding tools. One command. Auto-detected. Shareable.

---

pip install tokenmap
tokenmap

That's it. It reads your local data and renders a heatmap — plus an estimated cost breakdown — right in your terminal.

By default tokenmap includes your Claude Code data (and auto-detects any other installed tools), uses the dark-green theme, shows the cost breakdown, and stays terminal-only. Pass --export png (or svg/csv) when you want a shareable file.

Supported Tools

Tool	Data Source	What's Tracked
Claude Code	~/.claude/stats-cache.json	Tokens, models, sessions, costs
Codex CLI	~/.codex/sessions/*.jsonl	Tokens, models, session durations
OpenCode	~/.local/share/opencode/	Tokens, models, messages
Cursor	Cursor API + local state.vscdb	Tokens, models, usage events

tokenmap auto-detects which tools you have installed. No configuration needed.

Install

pip install tokenmap

Requires Python 3.10+.

System Dependencies

For PNG export, you need Cairo installed:

# macOS
brew install cairo

# Ubuntu/Debian
sudo apt install libcairo2-dev

# Fedora
sudo dnf install cairo-devel

Usage

# Basic — Claude + any other auto-detected tools, cost breakdown, terminal only
tokenmap

# Add your name to the heatmap
tokenmap --user yourname

# Pick tools (Claude is always included unless you pass --no-claude)
tokenmap --codex                 # Claude + Codex
tokenmap --no-claude             # everything detected EXCEPT Claude
tokenmap --no-claude --cursor    # Cursor only

# Filter by date
tokenmap --year 2025
tokenmap --since 2026-01-01 --until 2026-01-31
tokenmap --since 2026-05-01      # from a date through today

# Filter to one model (substring match)
tokenmap --model claude-opus-4-7
tokenmap --model opus            # every Opus variant

# Change the color theme (default is dark-green)
tokenmap --theme green

# Export a shareable file (off by default)
tokenmap --export png
tokenmap --export svg
tokenmap --export csv            # per-day token + cost rows
tokenmap --out ~/Desktop/my-ai-usage.png   # implies export, format from extension

# Copy PNG to clipboard (implies PNG export)
tokenmap --copy

# Dump raw stats as JSON (to stdout, or to a file with --out)
tokenmap --json
tokenmap --json --out stats.json

# Hide the cost breakdown (shown by default)
tokenmap --no-cost

# See all themes
tokenmap --list-themes

Themes

10 built-in themes — 5 light, 5 dark:

Dark	Light
dark-ember	green
dark-green (default)	purple
dark-purple	blue
dark-blue	amber
dark-mono	mono

Options

Flag	Description	Default
--user <name>	Username shown on the heatmap	—
--claude / --no-claude	Include Claude Code data	on
--codex	Also include Codex data	off
--opencode	Also include OpenCode data	off
--cursor	Also include Cursor data	off
--theme <name>	Color theme	dark-green
--export <fmt>	Write a file: png, svg, or csv	off (terminal only)
--no-export	Force terminal-only output (overrides --export/--out)	—
--out <path>	Custom output path (implies export; format from extension)	—
--copy	Copy PNG to clipboard (implies PNG export)	—
--year <year>	Filter to a specific year	last 365 days
--since <date>	Start date, inclusive (YYYY-MM-DD)	—
--until <date>	End date, inclusive (YYYY-MM-DD)	—
--model <name>	Filter to models matching this name (substring)	all models
--json	Output raw stats as JSON (to stdout, or --out file)	—
--cost / --no-cost	Show estimated cost breakdown by model	on
--list-themes	Show all available themes	—

Programmatic Usage

from tokenmap import aggregate_multi, render_terminal, render_svg, compute_stats
from tokenmap.types import RenderOptions

# Load data from all detected tools
panels = aggregate_multi()

# Render to terminal
render_terminal(panels, RenderOptions(theme="dark-green", user="myname"))

# Generate SVG
svg_string = render_svg(panels, RenderOptions(theme="dark-green", user="myname"))

# Access raw stats
for panel in panels:
    print(f"{panel.tool}: {panel.stats.total_tokens} tokens")

How It Works

tokenmap reads locally stored data from your AI coding tools. It never sends data anywhere — everything stays on your machine.

1. Detect — scans for installed tool data directories
2. Aggregate — merges token usage, sessions, and model stats across tools
3. Render — generates a terminal heatmap + cost breakdown
4. Export (optional) — with --export, saves a high-res PNG/SVG with stats panel, or a CSV of per-day tokens + cost

Privacy

• All data is read locally from your filesystem
• Nothing is uploaded or transmitted
• The only network request is Cursor's API (to fetch your own usage CSV, using your local auth token) — and even that's optional, with a local-only fallback that reads Cursor's state.vscdb on your machine when the API is unavailable.

Attribution

This project is a Python port of tokenviz by Harsh Kedia. Original source: https://github.com/harshkedia177/tokenviz

Licensed under MIT. Original copyright retained.

License

MIT