Quickstart

tokenolo wraps your coding agent behind a local proxy on 127.0.0.1. It removes redundant filler from your own messages before they're sent — removal-only, so the text that leaves is always a subsequence of what you wrote — and forwards everything else (system prompt, tools, tool output, files, history) straight to the model, untouched. Same agents, same commands.

This guide takes you from install to your first wrapped run.

Note: the CLI is moving fast. The commands below match the current onboarding flow — run tokenolo --help to confirm exact flags on your installed version.

1. Install the CLI

Install tokenolo globally with npm:

npm i -g tokenolo

Confirm it resolved:

tokenolo --version

2. Create an API key

Create an account at app.tokenolo.com/signup. The Free plan needs no card — it gives you 3 wraps/day at 20 requests/min, enough to run this guide.

Then generate an API key (it looks like tk_live_…) from the API keys page in your dashboard.

3. Sign in with your key

Authenticate the CLI against your account by passing the key inline:

tokenolo login --key tk_live_…

This links the CLI to your dashboard so runs count against your plan's daily quota.

4. Wrap your agent

Launch Claude Code through tokenolo. This starts a local proxy on 127.0.0.1 and points Claude Code at it, so the filler in each message is trimmed on the way to the model:

tokenolo wrap claude

Work exactly as you always do — type prompts, run tools, edit files. tokenolo trims redundant filler from each of your messages inline, forwards everything else untouched, then steps aside. No rewiring: it's a drop-in for Claude Code (and Codex — use tokenolo wrap codex).

What changes

The pipeline is prompt → agent → tokenolo wrap → filler removed → model. On every run:

• A modest token trim. Filler is removed from your messages before send — a small reduction, not a dramatic one.
• Private by design. Your code never leaves your machine; the request goes to the model under your own key, and nothing is stored.
• Cache-safe. Byte-stable edits keep your prompt cache valid across turns.

Your usage against the quota is visible in the dashboard.

Next steps

• Claude Code — wire tokenolo into your Claude Code workflow.
• How it works — see exactly what tokenolo trims and what it forwards untouched.