Measure the tax · trim it · keep the same quality
Claude Code Token Tuner — Cut Your Token Bill Without Losing Quality
Claude bills you for context size × turns. A bloated CLAUDE.md, auto-loaded rule files you forgot were on, and a few silent cache-busters quietly re-bill you every turn. The Token Tuner makes the tax visible, then shows you exactly how to trim it — same model, same skills, same quality.
Instant download · offline measuring tool · ~2-minute setup · no API key for the core
The problem
You can't trim what you can't see
Most people never measure their always-on token tax — so they pay for the same bytes on every turn of every session, forever.
▌ Flying blind
- A bloated CLAUDE.md re-loads on every single turn
- Auto-loaded rule files you forgot were even on
- Silent cache-busters re-bill your whole history at full price
- No number in front of you — so nothing ever gets fixed
▌ Token Tuner
- weigh_context.py counts your always-on tax — offline, no key
- A two-track Trim Playbook: paste-one-prompt beginner + advanced manual
- Names the silent cache-busters most token counters never show you
- Run it, trim, run it again, watch the number drop
What it does
Measure, trim, and dodge the silent spikes
Built from a real audit: a live setup measured down from ~13,900 to ~7,300 always-on tokens (−47%). Token counts are chars/4 estimates for before/after, not exact billing.
See the tax
weigh_context.py counts your always-on token tax. No API key, no tokens spent, runs offline. Run it, trim, run it again, watch the number drop.
Two tracks
BEGINNER: paste one prompt, do three free things — permissions.deny read-rules, a concise output style, one-task-per-session. ADVANCED: the deeper wins most people miss.
The hidden cache-busters
A mid-session model switch, the opusplan trap, and enabling fast mode mid-session all invalidate your prompt cache — so the next turn re-reads history without the cached prefix.
The new levers
Fork subagents that share the parent's cache prefix (up to ~90% less input on parallel work), a hook that shrinks a 10,000-line test log before it hits context, and Batch API at 50% off.
What's in the box
Everything to measure and trim
- ✓weigh_context.py — the measuring tool (offline, no key)
- ✓The Trim Playbook — beginner one-prompt track + advanced manual track
- ✓An illustrated setup guide + a full explainer with charts
- ✓A free audit-prompt you can share
- ✓One-command installer + a clean uninstaller
- ✓Fully yours to edit — change anything, keep it forever
⚠ Requirements — please read before buying
- Claude Code (this is a Claude Code optimization kit)
- macOS or Linux — Windows works via WSL
- Python 3 (already installed on Mac/Linux)
- No API keys for the core (one optional advanced tactic, Batch API, needs API billing)
Stop paying for the same bytes twice
Make the token tax visible — then cut it
Same model, same skills, same quality. You just stop re-paying for context you didn't need loaded. Unzip, install, read your number, trim, re-measure.
Instant download · 30-day Gumroad refund policy applies
Setting up Claude Code from scratch? Start with the Claude Code Setup Playbook — $29.
See all my Claude Code tools → expressive446.gumroad.com