waterprint

AI water consumption tracker for Claude Code. Turns tokens into liters — and puts that number in perspective.

💧 25.0 mL 🥄 5.00 teaspoons  •  Total: 25.0 mL 🥄 5.00 teaspoons
💧 300.0 mL 🥛 1.20 glasses of water  •  Total: 1.2 L 🍷 1.60 wine bottles

Shows session water (current conversation) and lifetime total (accumulated across all sessions).

Water rates are model-aware, based on Li et al. 2023 — "Making AI Less Thirsty". Claude 4.x rates are linear extrapolations — treat as estimates.

Install

npm install -g @syntaxerror-sys/water-print

Claude Code setup

Option 1 — Use waterprint as your status line

{
  "statusLine": {
    "type": "command",
    "command": "waterprint"
  }
}

Option 2 — Combine with your existing status line

If you already use a status line tool (e.g. ccstatusline), create a personal wrapper:

#!/bin/bash
INPUT=$(cat)
echo "$INPUT" | your-existing-statusline
echo "$INPUT" | waterprint

Save it anywhere (e.g. ~/.claude/statusline.sh), make it executable, and point settings.json to it.

Option 3 — Just try it without touching your config

waterprint --preview 50000

CLI

# Preview any token count
waterprint --preview 50000
# 💧 25.0 mL 🥄 5.00 teaspoons

# With a specific model
waterprint --preview 600000 --model claude-opus-4-8
# 💧 480.0 mL 🥛 1.92 glasses of water

# Override model when piping JSON
echo '<status-json>' | waterprint --model claude-opus-4-8

# Show usage history
waterprint --stats
# waterprint stats
#   Today:          0.000 mL  💧 0.000 water drops
#   7 days:        25.3 mL  🥄 5.06 teaspoons
#   Lifetime:     414.6 L  🍷 552.8 wine bottles
#   Sessions:  42

# Reset all accumulated stats
waterprint --reset

waterprint --version
waterprint --help

Supported models

| Model | mL / 1k tokens | Source | |---|---|---| | claude-haiku-4-5 | 0.30 | extrapolated | | claude-sonnet-4-6 | 0.50 | extrapolated | | claude-opus-4-8 | 0.80 | extrapolated | | claude-3-5-haiku | 0.28 | Li et al. 2023 | | claude-3-5-sonnet | 0.48 | Li et al. 2023 | | claude-3-opus | 0.75 | Li et al. 2023 |

Unknown models fall back to 0.50 mL / 1k tokens.

Comparisons

drops · teaspoons · shot glasses · glasses · water bottles · wine bottles · gallons · bathtubs · jacuzzis · Olympic pools · small lakes · oceans

Data

Session and lifetime stats are stored locally in ~/.waterprint/stats.json. Nothing is sent anywhere.

License

MIT