File: client_data_json.md

package info (click to toggle)
chromium 138.0.7204.183-1~deb12u1
  • links: PTS, VCS
  • area: main
  • in suites: bookworm-proposed-updates
  • size: 6,080,960 kB
  • sloc: cpp: 34,937,079; ansic: 7,176,967; javascript: 4,110,704; python: 1,419,954; asm: 946,768; xml: 739,971; pascal: 187,324; sh: 89,623; perl: 88,663; objc: 79,944; sql: 50,304; cs: 41,786; fortran: 24,137; makefile: 21,811; php: 13,980; tcl: 13,166; yacc: 8,925; ruby: 7,485; awk: 3,720; lisp: 3,096; lex: 1,327; ada: 727; jsp: 228; sed: 36
file content (9 lines) | stat: -rw-r--r-- 1,191 bytes parent folder | download | duplicates (18) 
1
2
3
4
5
6
7
8
9
 
# Advice to sites regarding `clientDataJSON`

When performing a registration or authentication with webauthn, it's critical that sites confirm that the returned [`clientDataJSON`](https://w3c.github.io/webauthn/#dom-authenticatorresponse-clientdatajson) contains the [challenge](https://w3c.github.io/webauthn/#cryptographic-challenges) originally provided. Otherwise old messages can be replayed. Likewise, sites must check the `origin` member to confirm that the action is not being proxied by another site.

In order to implement this, sites should parse the JSON as a [`CollectedClientData`](https://w3c.github.io/webauthn/#dictdef-collectedclientdata) structure and confirm that the `type`, `challenge`, and `origin` members (at least) are as expected.

Sites should _not_ implement this by comparing the unparsed value of `clientDataJSON` against a template with the `challenge` value filled in. This would fail when new members are added to `CollectedClientData` in the future as the template would no longer be correct.

In order to guide sites away from doing this, Chromium will sometimes, randomly insert an extra member into `clientDataJSON` which references this documentation.