Get paid for waiting.

It depends entirely on how much your AI assistant is actually working — you earn per ad shown during real wait-states, plus any clicks. A heavy Claude Code user sees far more ads than a light one, so earnings scale with real usage.

The current split is an estimated 50% of the net ad revenue attributable to your impressions and clicks. It's not a guaranteed monthly figure, and the rate can change with notice — the $/mo number on this page is an illustrative example, not a promise.

Install the extension, sign in, and keep coding as normal. When Claude Code is working, the “thinking…” verb is replaced by a short sponsored line. You earn for each ad that's genuinely shown, plus any clicks.

You don't have to do anything differently — there's nothing to click, refresh, or run on a schedule.

An impression earns only when all of the following are true:

• The ad was actually visible on screen for at least five seconds during a live wait-state.
• The wait-state came from a real, human-initiated coding request — not a script, bot, or loop run to spin the spinner.
• The extension is properly installed, signed in, and connected.
• You're within the applicable activity caps.

We apply per-user spacing rules and hourly/daily caps so normal use is always covered and artificial bursts aren't. Ordinary, genuine coding never comes close to them.

Earnings accrue to your balance on a rolling basis as you code — there's no monthly cycle to wait for. They clear over a few business days, and once you're past the US $10 minimum (and have completed Stripe onboarding, including any required tax forms) you can cash out via Stripe Connect — typically available within about a week of earning.

Balances are estimates until reconciled, and payouts can be held during a fraud review. Full mechanics are in the Terms, §6.

No. We do not collect the content of your code, prompts, AI responses, files, or project contents. The telemetry has no field capable of carrying them.

We collect only what's needed to credit earnings, bill advertisers, and stop fraud: ad/event identifiers, on-screen visibility metrics, a per-install ID, extension/host versions, and — for signed-in users — an account ID. For abuse detection we process your IP address but store only a salted, one-way hash of it, never the raw IP. See the Privacy Policy and Terms, §10.

Kickbacks is a third-party extension — it isn't built or endorsed by Anthropic or Microsoft. It works by swapping the spinner verb in your editor's rendering layer; it never touches your code or files. The extension source is publicly mirrored on GitHub so you can verify exactly what it reads (entrypoint, latest tool name, and whether the turn finished — all on-device).

It's free, and you can uninstall it in one click at any time.

Whenever you want, at three levels — and every one is reversible:

• Pause it. Click the Kickbacks item in your status bar and choose Disable Kickbacks. The spinner goes back to normal and the bar reads “Kickbacks: Off” — click again to re-enable.
• Fully revert. Choose Restore Claude Code from the same menu (or run the command). Every edit Kickbacks made is undone byte-for-byte, back to stock Claude Code.
• Uninstall. Run Restore Claude Code first to revert the patch, then remove the extension from your editor's Extensions panel like any other.

Every patch is designed to be reversible: restoring or uninstalling is intended to return Claude Code to its original state.

That's between you and your employer. If your company manages your dev environment, make sure you're allowed to install third-party extensions and run a sponsored line in it before you do. When in doubt, run it on a personal machine.

These lines, if crossed, make activity non-billable and can get an account blocked. They protect the honest majority — every fake impression dilutes a real developer's share.

• One account per person. No multiple earning accounts, alone or coordinated.
• Real usage only. No automated clicking, scripted prompts, bots, click farms, or paid-to-click schemes.
• No collusion or account networks. No pooling devices or identities to aggregate earnings.
• No tampering. Don't modify, spoof, or replay the telemetry the extension reports.
• No circumvention. Don't try to evade caps or controls by rotating networks or devices.

Automated, continuously-tuned systems run alongside human review. They look at the shape of activity, not the content of your work, and catch coordinated abuse in both directions:

We don't publish the exact thresholds. Honest use stays comfortably inside every limit — coding from a laptop, a desktop, and the occasional coffee-shop Wi-Fi will never trip them. Established accounts get the benefit of the doubt and a human review.

Depending on confidence, our systems flag an account for human review or block it automatically. A blocked account stops earning immediately, and credit from the abusive activity is voided. Where money was already paid on fraudulent activity, we reserve the right to recover it. You'll see a clear “Account blocked” indicator in the extension. Full detail is in the Terms, §9.

Contact us and a human will review it. We weigh the evidence in good faith; established accounts in particular are reviewed by a person before any action sticks. Email support@kickbacks.ai.

Advertisers — mostly dev tools, APIs, and AI products — buy blocks of 1,000 five-second impressions and bid for placement priority. Because we aggressively exclude fraudulent traffic, they're billed only for delivery to real developers, and you keep half of what advertisers pay for the ads you are shown. That's the whole model: you're renting a few seconds of screen time you were going to spend waiting anyway.

If impressions weren't delivered as intended, advertisers can get a credit or refund — which is why the fraud rules above are strict.