How to Decode a JWT Token Safely

A decoded JWT can be readable without being trustworthy. Inspect only synthetic or non-sensitive test data in an online decoder.

Understand the three parts

To decode a typical signed compact JWT, split its three dot-separated segments and Base64url-decode the header and payload as JSON. The third part is a signature or message authentication code. Reading the first two parts does not validate the third. Some JWTs are encrypted and use a different compact structure; this decoder is intended for readable three-part tokens.

The header may identify an algorithm and token type. The payload contains claims about the token. Claims are assertions, not facts until the token has been properly verified.

Try a deliberately fake demonstration token

Use this fabricated example only. Its third segment is a placeholder, not a valid signature. It is not issued by any service and must never be accepted for authentication:

eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJkZW1vIn0.ZmFrZQ

The header decodes to {"alg":"HS256"} and the payload to {"sub":"demo"}. The free JWT Decoder displays decoded JSON as you type or paste. It cannot establish that the token was signed using HS256 or that demo is a real authenticated subject.

Try JWT Decoder

If the preview does not load, open JWT Decoder and customize it.

Read time and identity claims

Common registered claims in RFC 7519 include iss (issuer), sub (subject), aud (audience), iat (issued at), nbf (not before) and exp (expiration). Time claims use NumericDate values in seconds, not milliseconds. The widget formats supported time claims and can show an expiry indicator when an expiration is present.

An unexpired-looking token may be forged, have the wrong issuer or audience, or be revoked. A decoder's apparent active badge is only an interpretation of its unverified time claim, never an authorization decision.

Verify at the trusted application boundary

Use an established JWT library and your application's intended algorithm allow-list, trusted issuer keys and validation rules. Verify the signature and the required claims, including expiry, issuer and audience where applicable. Follow JWT Best Current Practices (RFC 8725). Never select trust or authorization based only on values supplied by an unverified token.

The widget does not verify a signature, fetch issuer keys, decrypt tokens or validate an account. Use the proper server-side application verification flow for authentication.

Never paste live credentials into an online tool

Real access tokens, refresh tokens and session cookies can grant access. Avoid pasting them into online decoders, shared Notion pages, generated URLs, logs, screenshots or chat messages. The widget decodes in the browser, but the hosted page and analytics may make network requests; that is not a blanket privacy assurance. For real troubleshooting, use an approved local tool on your own machine and sanitize the output before sharing.

Customize only appearance and size if you embed the widget with practice data. C copies the decoded payload and X clears the field when keyboard focus is outside the input; clipboard permissions may block copying. Clearing the UI does not revoke a token. If a real credential was exposed, follow your organization's revocation or incident process.

Common questions

Can anyone read an ordinary signed JWT payload?

Yes, the encoded header and payload are not encrypted by signing. Do not put secrets in readable claims.

Is decoding the same as verifying?

No. Decoding reads the data. Verification establishes authenticity and required claims under a trusted policy.

Can I paste my production access token here?

No. Use an approved local environment with production credentials and use only fabricated test data in this online example.