github-starred/better-auth
2
0
Fork 0
mirror of https://github.com/better-auth/better-auth.git synced 2026-05-11 23:41:17 -05:00
Code Issues 2.5k Packages Projects Releases 112 Wiki Activity

[GH-ISSUE #3226] Invalid keyPairConfig options available to user #9529

New Issue
Closed
opened 2026-04-13 05:01:37 -05:00 by GiteaMirror · 2 comments
GiteaMirror commented 2026-04-13 05:01:37 -05:00
Owner
Copy Link

Originally created by @LightTab2 on GitHub (Jun 30, 2025).
Original GitHub issue: https://github.com/better-auth/better-auth/issues/3226

Is this suited for github?

  • Yes, this is suited for github

To Reproduce

Pass

 {
	alg: "EdDSA",
	crv: "Ed448",
}

or

 {
	alg: "ECDH-ES",
	crv: any of ["P-256" | "P-384" | "P-521"],
}

to

plugins: [
	jwt({
		jwks: {
			keyPairConfig: {
				...
			},
		}),
	],

Current vs. Expected behavior

Expected:

These options should not be allowed

Current:

 {
	alg: "EdDSA",
	crv: "Ed448",
}

uses "Ed25519" instead.

 {
	alg: "ECDH-ES",
	crv: any of ["P-256" | "P-384" | "P-521"],
}

cannot be used to sign tokens.

What version of Better Auth are you using?

1.2.12

Provide environment information

Irrelevant

Which area(s) are affected? (Select all that apply)

Backend

Auth config (if applicable)

Additional context

According to JOSE Algorithm Key Requirements EdDSA algorithm needs to be used with Ed25519 curve.

These bugs were found while writing tests for PR #3219 solving another issue #3218 , related comment:

Tests concerning 'ECDH-ES' are also failing.

I inspected the private keys and they have no 'sing' usage assigned to them:

CryptoKey {
  type: 'private',
  extractable: false,
  algorithm: { name: 'ECDH', namedCurve: 'P-256' },
  usages: [ 'deriveBits' ]
}

example of another key:

CryptoKey {
  type: 'private',
  extractable: false,
  algorithm: {
    name: 'RSASSA-PKCS1-v1_5',
    modulusLength: 2048,
    publicExponent: [Uint8Array],
    hash: [Object]
  },
  usages: [ 'sign' ]
}

This one has that field.

It is apparently pre-Diffie-Hellman, we would need to use deriveBits to create a common secret and then an actual key for signing.
I will drop test support for this case (not sure if that's a valid move eww)

Originally posted by @LightTab2 in https://github.com/better-auth/better-auth/issues/3219#issuecomment-3017310172

Originally created by @LightTab2 on GitHub (Jun 30, 2025). Original GitHub issue: https://github.com/better-auth/better-auth/issues/3226 ### Is this suited for github? - [x] Yes, this is suited for github ### To Reproduce Pass ```js { alg: "EdDSA", crv: "Ed448", } ``` or ```js { alg: "ECDH-ES", crv: any of ["P-256" | "P-384" | "P-521"], } ``` to ```js plugins: [ jwt({ jwks: { keyPairConfig: { ... }, }), ], ``` ### Current vs. Expected behavior ### Expected: These options should not be allowed ### Current: ```js { alg: "EdDSA", crv: "Ed448", } ``` uses "**Ed25519**" instead. ```js { alg: "ECDH-ES", crv: any of ["P-256" | "P-384" | "P-521"], } ``` cannot be used to sign tokens. ### What version of Better Auth are you using? 1.2.12 ### Provide environment information ```bash Irrelevant ``` ### Which area(s) are affected? (Select all that apply) Backend ### Auth config (if applicable) ```typescript ``` ### Additional context According to JOSE [Algorithm Key Requirements](https://github.com/panva/jose/issues/210) **EdDSA** algorithm needs to be used with **Ed25519** curve. These bugs were found while writing tests for PR #3219 solving another issue #3218 , related comment: > Tests concerning 'ECDH-ES' are also failing. > > I inspected the **private keys** and they have no 'sing' usage assigned to them: ```js CryptoKey { type: 'private', extractable: false, algorithm: { name: 'ECDH', namedCurve: 'P-256' }, usages: [ 'deriveBits' ] } ``` > example of another key: ```js CryptoKey { type: 'private', extractable: false, algorithm: { name: 'RSASSA-PKCS1-v1_5', modulusLength: 2048, publicExponent: [Uint8Array], hash: [Object] }, usages: [ 'sign' ] } ``` > This one has that field. > > It is apparently pre-Diffie-Hellman, we would need to use deriveBits to create a common secret and then an actual key for signing. > I will drop test support for this case (not sure if that's a valid move eww) _Originally posted by @LightTab2 in https://github.com/better-auth/better-auth/issues/3219#issuecomment-3017310172_
GiteaMirror added the lockedbug labels 2026-04-13 05:01:37 -05:00
GiteaMirror commented 2026-04-13 05:01:39 -05:00
Author
Owner
Copy Link

@dosubot[bot] commented on GitHub (Jun 30, 2025):

You're correct—Better Auth's jwt plugin currently allows passing keyPairConfig options like { alg: "EdDSA", crv: "Ed448" } or { alg: "ECDH-ES", crv: "P-256" }, but these are not properly validated. For EdDSA, only Ed25519 is supported for signing in JOSE; if you specify Ed448, the implementation silently falls back to Ed25519 instead of throwing an error. For ECDH-ES, it's a key agreement algorithm, not a signing algorithm, so keys generated this way can't be used to sign tokens—they only have deriveBits usage, not sign usage, which matches what you observed in your tests. There is no comprehensive validation in the codebase to prevent these mismatches, so invalid configs can slip through and either fail silently or just not work as expected (source).

The correct config for signing JWTs is { alg: "EdDSA", crv: "Ed25519" }, or you can use RSA/EC algorithms like RS256, ES256, or ES512 as documented. ECDH-ES should not be used for signing. If you want stricter validation, you’d need to add explicit checks for allowed algorithm/curve combinations in your own code or propose a validation enhancement upstream.

Your plan to drop test support for ECDH-ES signing is correct—it's not a valid use case. If you need references, see the jwt plugin code and the JOSE algorithm requirements.

To reply, just mention @dosu.

How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other  Join Discord Share on X

<!-- gh-comment-id:3019124349 --> @dosubot[bot] commented on GitHub (Jun 30, 2025): <!-- Answer --> You're correct—Better Auth's jwt plugin currently allows passing keyPairConfig options like `{ alg: "EdDSA", crv: "Ed448" }` or `{ alg: "ECDH-ES", crv: "P-256" }`, but these are not properly validated. For EdDSA, only Ed25519 is supported for signing in JOSE; if you specify Ed448, the implementation silently falls back to Ed25519 instead of throwing an error. For ECDH-ES, it's a key agreement algorithm, not a signing algorithm, so keys generated this way can't be used to sign tokens—they only have `deriveBits` usage, not `sign` usage, which matches what you observed in your tests. There is no comprehensive validation in the codebase to prevent these mismatches, so invalid configs can slip through and either fail silently or just not work as expected ([source](https://github.com/better-auth/better-auth/blob/7c4719ed77941bc1a685dc16550be10ad980f3df/packages/better-auth/src/plugins/jwt/index.ts)). The correct config for signing JWTs is `{ alg: "EdDSA", crv: "Ed25519" }`, or you can use RSA/EC algorithms like RS256, ES256, or ES512 as documented. ECDH-ES should not be used for signing. If you want stricter validation, you’d need to add explicit checks for allowed algorithm/curve combinations in your own code or propose a validation enhancement upstream. Your plan to drop test support for ECDH-ES signing is correct—it's not a valid use case. If you need references, see the [jwt plugin code](https://github.com/better-auth/better-auth/blob/7c4719ed77941bc1a685dc16550be10ad980f3df/packages/better-auth/src/plugins/jwt/index.ts) and the [JOSE algorithm requirements](https://github.com/panva/jose/issues/210). <!-- Dosu Comment Footer --> *To reply, just mention [@dosu](https://go.dosu.dev/dosubot).* --- <sup>How did I do? [Good](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=great_response) | [Irrelevant](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=irrelevant_answer) | [Incorrect](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=incorrect_sources) | [Verbose](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=too_verbose) | [Hallucination](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=hallucination) | [Report 🐛](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=bug_report) | [Other](https://app.dosu.dev/response-feedback/4b663c4b-15c5-4363-9f8e-dea4e5cfefb2?feedback_type=other)</sup>&nbsp;&nbsp;[![Join Discord](https://img.shields.io/badge/join-5865F2?logo=discord&logoColor=white&label=)](https://go.dosu.dev/discord-bot)&nbsp;[![Share on X](https://img.shields.io/badge/X-share-black)](https://twitter.com/intent/tweet?text=%40dosu_ai%20helped%20me%20solve%20this%20issue!&url=https%3A//github.com/better-auth/better-auth/issues/3226)
GiteaMirror commented 2026-04-13 05:01:39 -05:00
Author
Owner
Copy Link

@LightTab2 commented on GitHub (Sep 14, 2025):

It was fixed by #4161

<!-- gh-comment-id:3289056832 --> @LightTab2 commented on GitHub (Sep 14, 2025): It was fixed by #4161
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
adapter
astro
awaiting external contributor
blocked
breaking
breaking change
bug
c-devops
core
credentials
database
dependencies
devops
devtools
docs
documentation
duplicate
elysia
enhancement
enterprise
expo
express
fastify
good first issue
help wanted
hono
identity
infra
integration
invalid
javascript
locked
maintenance
need-more-information
needs: info
needs: repro
nextjs
nuxt
oauth
organization
P0
payments
perf
platform
plugin
pull-request
Mirrored from GitHub Pull Request
 question
ready
regression
remix
security
social-provider
solid
stale
svelte
tanstack-start
tracking
version-bump
vue
wontfix
No Label bug locked
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/better-auth#9529
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?