Cloud consoles, payment dashboards and AI playgrounds all show secrets that let someone else spend your money or read your data. Learn where API keys leak while you browse, how to keep tokens out of the browser, and how to rotate keys so a leak does less damage.
Pages, extensions, the clipboard and your screen can all see what you paste.
An extension that can "read and change all your data on all websites" can read every field on a dashboard page, including a key you reveal or paste. Keep the list short and remove extensions you no longer use.
Browser autofill is built for login forms. A key saved as a password can be offered on the wrong field, and it syncs to every device signed in to that browser profile.
Windows clipboard history (Win+V) and clipboard managers keep copied text. A key you paste once can stay in that history.
A key revealed on a shared screen, in a recorded demo or in a bug-report screenshot is a leaked key. Reveal secrets only when nobody can see.
Any script on the same origin can read localStorage and sessionStorage. One cross-site scripting (XSS) bug, or one compromised third-party script, exposes every token stored there.
A phishing copy of a cloud console collects a key as easily as a password. Open dashboards from bookmarks and check the domain.
The safest key never passes through a browser tab. When you must use a dashboard, limit how often a long-lived secret appears on screen and what it can do if it escapes.
Rotation limits how long a leaked key stays useful. An overlap rotation replaces a key without breaking the service.
Create a new key with the same or narrower permissions.
Put the new key in your secrets manager and redeploy every service that uses it.
Check the provider's logs: traffic uses the new key and the old key is idle.
Delete or disable the old key. Write down the date so you know when the next rotation is due.
Get a key's age, its next rotation date, and how long it would keep working if it leaked today.
How it works: next rotation = last rotation + interval. If nobody notices a leak, the key keeps working until that date, so the days left are the exposure window. 90 days or less is the interval in the CIS Benchmarks for AWS and Google Cloud and in AWS Security Hub control IAM.3. The planner runs only in your browser. Never type a key into it or into any other web page.
Why password managers struggle with API key fields, how browser storage options compare, and a checklist for auto-fill tools.
An end-to-end encrypted password manager is much safer than a text file, a notes app or a chat message. Store keys as secure notes rather than as logins, so the manager does not offer them for autofill on unrelated fields.
90 days or less is the interval used by the CIS Benchmarks for AWS and Google Cloud and by AWS Security Hub control IAM.3. Rotate at once after a suspected leak or when someone with access leaves.
Revoke or rotate the key first. Then check the provider's usage logs for calls you did not make, and only then clean up the place where it leaked. Deleting a commit or a message does not make a leaked key safe again.