Privacy

Does claudearn read my code?

Published June 25, 2026 · claudearn

If you're considering an extension that lives inside your editor and pays you, "does claudearn read my code?" is exactly the right question to ask first. You've seen a thousand "earn crypto" pitches and you're rightly suspicious. So here's the honest answer up front, then the architecture that backs it up — because privacy you can't verify is just a pinky-promise with better fonts.

Read what the extension actually does →See how it works

The short answer: no

No. claudearn does not read your code, your prompts, your open files, your repo, or anything else on your machine. The extension does exactly two things: it renders one sponsored line in your AI agent's thinking spinner, and it reports that a view happened. That is the entire list of things it does to your machine.

It can't read your code because it was never built to. There is no file access in the data path, no prompt capture, no telemetry that ships your work anywhere. The thing it sends home is a count, not your codebase.

What the extension actually does: renders ads, reports views

Think of it as a tiny billboard renderer wired into the wait-state. While your agent is "Thinking…", claudearn slots one tasteful sponsored line into that spinner for about five seconds. When the line is shown, the extension reports a single event — an impression occurred — so your earnings can be tallied and paid out.

That's the whole loop. The ad line is fetched, displayed, and counted. None of those steps require — or have access to — what you're actually coding.

  • ▸Renders one sponsored line in the spinner — never blocking, one-click reversible, opt-in.
  • ▸Reports a view event so impressions can be counted and paid.
  • ▸Does not read, capture, transmit, or store your code, prompts, files, or repo contents.

Privacy by architecture, not a pinky-promise

"We never see your data" is the kind of vague reassurance every sketchy tool offers. We'd rather show you the shape of the system than ask you to trust a sentence. claudearn is private because of how it's built, not because we'd feel bad otherwise.

The earning side and the content you write never touch. Your agent does its thinking locally, as it always has. claudearn only operates on the spinner's display layer — it paints a line and counts that the line was seen. The advertiser pays for an impression, so an impression is the only thing that needs to leave your machine. There's no business reason to read your code, and no code path that does it.

What leaves your machine and what never does

It helps to be concrete about the boundary. Here's the split, plainly:

  • ▸Leaves your machine: a view event (an impression was rendered) so earnings can be counted.
  • ▸Never leaves your machine: your source code, your prompts to the agent, your file names, your repo contents, your keystrokes.
  • ▸Where the money lives: payouts settle on Solana mainnet in USDC — real money, claimable to any wallet such as Phantom, and verifiable on-chain via Solscan.

Is claudearn safe — and is claudearn legit?

Safe in the sense that matters here: it doesn't exfiltrate your work, it doesn't block your workflow, and you can turn it off in one click. It's free for VS Code, Claude Code, and Codex (the terminal CLI), and it's opt-in — nothing renders until you say so. This is the non-tracking model done literally: there's nothing to track because the code path was never given your work to look at.

Legit in the financial sense too. Advertisers bid in a live market in blocks of 1,000 impressions; the top bid takes the #1 spinner slot. At roughly a $20 CPM, about $10 of each 1,000 impressions reaches the developer — roughly half. Honest expectation: about $15–$60 a month, and it varies with advertiser demand and how many hours you actually code. It's beer money for seconds you were already losing, not a salary, and we won't pretend otherwise.

How to verify it yourself

You don't have to take our word for any of this. Treat it the way you'd treat any dependency you let near your editor.

Inspect what the extension requests and sends — you'll find ad content coming in and view events going out, and nothing resembling your files. Read the terms on our legal page for the formal version of the same boundary. And when you claim earnings (the minimum is $1.00), follow the payout on Solscan: the USDC landing in your wallet is a public, on-chain fact, not a screenshot we made up.

  • ▸Watch the network: ad content in, a view event out — no file or prompt payloads.
  • ▸Read the boundary in writing on the legal page.
  • ▸Trace any payout on-chain via Solscan once you pass the $1.00 minimum.

Install with confidence

If the only thing standing between you and a paid spinner was "does this thing read my code" — now you have the answer and a way to check it. The extension is free, opt-in, and reversible in one click. Flip it on, keep coding, and let the seconds you were already waiting earn their keep.

Frequently asked questions

Does claudearn read my code or prompts?

No. The extension renders one sponsored line in the agent's spinner and reports that the view happened. It has no access path to your source code, prompts, files, or repo — that data never leaves your machine.

What information actually leaves my machine?

Only a view event — a count that an impression was rendered — so your earnings can be tallied and paid. Your code, prompts, and file contents stay local.

Is claudearn legit, and where does the money come from?

Yes. Advertisers bid for the #1 spinner slot in blocks of 1,000 impressions. At roughly a $20 CPM, about $10 per 1,000 impressions reaches you — paid in USDC on Solana mainnet, claimable to any wallet and verifiable on Solscan. Realistic earnings are about $15–$60/mo and vary with demand and coding hours.

Can I turn it off?

Yes. It's opt-in and one-click reversible. The sponsored line is never blocking and shows for about five seconds during the wait-state you were already sitting through.