tokentrim 55‑70% cost savings
tokentrim can reduce your Claude Code costs by roughly 55‑70 %. Install the free toolkit, enable the compaction window, and watch your token usage drop.
How it works for a single Claude Code project
- Clone the repo or add the plugin to your environment.
- Enable the “compaction window” hook (only in the full install).
- Run your code as usual; tokentrim trims tool output, reads files in ranges, and spools long output to disk.
- Use the
replaycommand to see your own savings numbers.
What you get and the price
- Free tier – all hooks, three strength profiles, replay command, local dashboard, daily‑budget guard, cold‑cache alert. Nothing leaves your machine.
- Pro lifetime – $49 one‑time (first 500 buyers). Adds a hosted dashboard usable on any device, Monday digest, weekly budget alert at 80 %, Slack totals, and support for up to three machines.
- Pro monthly – $9 per month with the same Pro features.
All features require Python 3.9+ and run on macOS or Linux.
FAQ
How much history does tokentrim actually skip? On the author’s 135 real conversations, the compaction window reduced history re‑read by 75 % and extended token limits by about 4.0 ×.
Do I need to send any data to a server? No. By default nothing leaves your machine. The hosted dashboard in Pro is optional.
Can I see the savings for my own runs? Yes. The replay command displays your personal cost‑reduction numbers after each session.
Updated 2026-11-02