Free: see where your Claude Code token spend goes + the tactics that cut it

README.md

Claude Code Token-Saver — FREE sample

Post this publicly (GitHub gist / repo README / Reddit / Show HN). It's the runnable, free-to-try hook that makes Show HN and Reddit value-first posts legitimate. The full kit (18 tactics + drop-in configs + rules + scripts) is the paid upgrade.

Free: the 5 tactics that are ~80% of the savings

1. Route the model to the task. Haiku for mechanical edits/scaffolding, Sonnet for daily feature work, Opus only for genuinely hard reasoning. Pinning everything to the top model is the most expensive default there is.
2. /clear between unrelated tasks. Every turn re-sends the whole conversation as input tokens; don't pay to drag a finished task into the next one.
3. Stop re-reading files you just edited. The edit already confirmed success — re-reading to "verify" doubles the cost for zero new info.
4. (API) Turn on prompt caching + the Batch API. Cached input and non-realtime batch jobs are dramatically cheaper. Most people never switch them on.
5. Cap extended thinking on routine sessions. Great for hard problems, pure overhead on easy ones.

Free: the token-report script

A read-only script (Windows PowerShell + bash) that scans your local Claude Code transcripts and shows input/output/cached tokens + cache-hit % over the last N days. Run it weekly; watch the number drop. (Included in this sample.)

Want the full thing?

The paid Claude Code Token-Saver Kit adds:

• The full 18-tactic playbook, ranked by $ saved per minute
• A ready-to-merge cost-control settings.json (model default, thinking cap, hygiene hook)
• A model-routing decision table
• A drop-in cost-control.md rules file
• Single-seat commercial license + 14-day refund

👉 https://payhip.com/b/QlJE9

Prices and model names change — always confirm current numbers in Anthropic's official docs. The tactics are durable.

token-report.ps1

# token-report.ps1 — quick estimate of recent Claude Code usage (Windows PowerShell)
# Reads Claude Code project transcripts and tallies approximate token usage so you
# can watch your spend trend down week over week. Read-only; touches nothing else.
#
# Usage:  powershell -ExecutionPolicy Bypass -File token-report.ps1
#         powershell -File token-report.ps1 -Days 7

param(
    [int]$Days = 7
)

$projectsDir = Join-Path $env:USERPROFILE ".claude\projects"

if (-not (Test-Path $projectsDir)) {
    Write-Host "No Claude Code projects directory found at $projectsDir" -ForegroundColor Yellow
    exit 0
}

$cutoff = (Get-Date).AddDays(-$Days)
$files = Get-ChildItem -Path $projectsDir -Recurse -Filter *.jsonl -ErrorAction SilentlyContinue |
    Where-Object { $_.LastWriteTime -ge $cutoff }

if (-not $files) {
    Write-Host "No transcript activity in the last $Days day(s)." -ForegroundColor Yellow
    exit 0
}

$inputTokens  = 0
$outputTokens = 0
$cacheRead    = 0

foreach ($f in $files) {
    foreach ($line in [System.IO.File]::ReadLines($f.FullName)) {
        if ($line -notmatch '"usage"') { continue }
        try { $obj = $line | ConvertFrom-Json -ErrorAction Stop } catch { continue }
        $u = $obj.message.usage
        if ($null -eq $u) { $u = $obj.usage }
        if ($null -eq $u) { continue }
        if ($u.input_tokens)               { $inputTokens  += [int]$u.input_tokens }
        if ($u.output_tokens)              { $outputTokens += [int]$u.output_tokens }
        if ($u.cache_read_input_tokens)    { $cacheRead    += [int]$u.cache_read_input_tokens }
    }
}

$totalIn = $inputTokens + $cacheRead
$cachePct = if ($totalIn -gt 0) { [math]::Round(($cacheRead / $totalIn) * 100, 1) } else { 0 }

Write-Host ""
Write-Host "Claude Code usage - last $Days day(s)" -ForegroundColor Cyan
Write-Host "----------------------------------------"
Write-Host ("Input tokens (fresh):  " + $inputTokens.ToString("N0"))
Write-Host ("Input tokens (cached): " + $cacheRead.ToString("N0") + "  " + $cachePct + "% from cache")
Write-Host ("Output tokens:         " + $outputTokens.ToString("N0"))
Write-Host ""
Write-Host "Tip: higher cache % = lower bill. If cache % is near 0 and you build on the API," -ForegroundColor DarkGray
Write-Host "see GUIDE.md tactic #2 (prompt caching)." -ForegroundColor DarkGray

token-report.sh

#!/usr/bin/env bash
# token-report.sh — quick estimate of recent Claude Code usage (macOS/Linux)
# Reads Claude Code project transcripts and tallies approximate token usage.
# Read-only. Requires: jq.
#
# Usage:  ./token-report.sh [DAYS]   (default 7)

set -euo pipefail
DAYS="${1:-7}"
DIR="$HOME/.claude/projects"

if ! command -v jq >/dev/null 2>&1; then
  echo "This script needs 'jq'. Install it (brew install jq / apt-get install jq) and retry." >&2
  exit 1
fi

if [ ! -d "$DIR" ]; then
  echo "No Claude Code projects directory at $DIR"
  exit 0
fi

in_fresh=0; in_cache=0; out=0

while IFS= read -r -d '' f; do
  while IFS= read -r line; do
    case "$line" in
      *'"usage"'*) ;;
      *) continue ;;
    esac
    i=$(printf '%s' "$line" | jq -r '(.message.usage // .usage).input_tokens // 0' 2>/dev/null || echo 0)
    c=$(printf '%s' "$line" | jq -r '(.message.usage // .usage).cache_read_input_tokens // 0' 2>/dev/null || echo 0)
    o=$(printf '%s' "$line" | jq -r '(.message.usage // .usage).output_tokens // 0' 2>/dev/null || echo 0)
    in_fresh=$(( in_fresh + i ))
    in_cache=$(( in_cache + c ))
    out=$(( out + o ))
  done < "$f"
done < <(find "$DIR" -name '*.jsonl' -type f -mtime "-${DAYS}" -print0)

total_in=$(( in_fresh + in_cache ))
pct=0
[ "$total_in" -gt 0 ] && pct=$(( in_cache * 100 / total_in ))

echo ""
echo "Claude Code usage - last ${DAYS} day(s)"
echo "----------------------------------------"
printf "Input tokens (fresh):  %'d\n"  "$in_fresh"  2>/dev/null || echo "Input tokens (fresh):  $in_fresh"
printf "Input tokens (cached): %'d  (%s%% from cache)\n" "$in_cache" "$pct" 2>/dev/null || echo "Input tokens (cached): $in_cache (${pct}% from cache)"
printf "Output tokens:         %'d\n"  "$out"        2>/dev/null || echo "Output tokens:         $out"
echo ""
echo "Higher cache % = lower bill. See GUIDE.md tactic #2 if it's near 0."