github-starred/better-auth
2
0
Fork 0
mirror of https://github.com/better-auth/better-auth.git synced 2026-05-24 08:01:56 -05:00
Code Issues 2.5k Packages Projects Releases 112 Wiki Activity

[GH-ISSUE #8335] oauth-provider: OIDC Discovery and JWT signing conformance gaps #28382

New Issue
Open
opened 2026-04-17 19:49:40 -05:00 by GiteaMirror · 1 comment
GiteaMirror commented 2026-04-17 19:49:40 -05:00
Owner
Copy Link

Originally created by @gustavovalverde on GitHub (Mar 3, 2026).
Original GitHub issue: https://github.com/better-auth/better-auth/issues/8335

Originally assigned to: @gustavovalverde on GitHub.

Summary

The @better-auth/oauth-provider plugin combined with the jwt() plugin has several deviations from the OIDC specifications that prevent relying parties from integrating with standard OAuth libraries without workarounds.

1. Discovery metadata omits RS256 (OIDC Discovery 1.0 violation)

oidcServerMetadata in metadata.ts derives id_token_signing_alg_values_supported from keyPairConfig.alg, producing a single-element array. The jwt plugin defaults to EdDSA, so the discovery document advertises ["EdDSA"].

OIDC Discovery 1.0 states:

id_token_signing_alg_values_supported — REQUIRED. [...] The algorithm RS256 MUST be included.

Any deployment using the default jwt plugin config is non-conformant.

Consequence: OIDC conformance test suites fail. Standard RP libraries (openid-client, next-auth, Spring Security) that default to RS256 for ID Token verification reject tokens signed with an unadvertised algorithm.

2. No per-client id_token_signed_response_alg

The oauthClient schema has no id_token_signed_response_alg field. OIDC Dynamic Client Registration 1.0 defines this as standard client metadata with a default of RS256.

Consequence: Clients cannot negotiate their preferred signing algorithm during registration. Providers that support multiple algorithms (e.g., RS256 + EdDSA) have no mechanism to sign per-client.

3. Discovery only supports advertising a single algorithm

oidcServerMetadata reads one alg from keyPairConfig and wraps it in an array. There is no way to advertise multiple supported algorithms (e.g., ["RS256", "EdDSA"]), even though OIDC Discovery defines id_token_signing_alg_values_supported as an array for this purpose.

4. Custom sign requires remoteUrl, which disables built-in JWKS

The jwt plugin enforces: jwt.sign → requires jwks.remoteUrl → disables the built-in /jwks endpoint (returns 404). This means adding a secondary signing algorithm (e.g., a post-quantum algorithm alongside EdDSA) forces the developer to take over the entire key management and JWKS serving stack, even for the standard algorithm that the built-in path already handles.

Consequence: There is no way to extend the signing capabilities incrementally. Any custom algorithm requires replacing the entire JWT infrastructure.

Spec references

Originally created by @gustavovalverde on GitHub (Mar 3, 2026). Original GitHub issue: https://github.com/better-auth/better-auth/issues/8335 Originally assigned to: @gustavovalverde on GitHub. ## Summary The `@better-auth/oauth-provider` plugin combined with the `jwt()` plugin has several deviations from the OIDC specifications that prevent relying parties from integrating with standard OAuth libraries without workarounds. ## 1. Discovery metadata omits RS256 (OIDC Discovery 1.0 violation) `oidcServerMetadata` in `metadata.ts` derives `id_token_signing_alg_values_supported` from `keyPairConfig.alg`, producing a single-element array. The jwt plugin defaults to `EdDSA`, so the discovery document advertises `["EdDSA"]`. [OIDC Discovery 1.0](https://openid.net/specs/openid-connect-discovery-1_0.html) states: > `id_token_signing_alg_values_supported` — REQUIRED. [...] **The algorithm RS256 MUST be included.** Any deployment using the default jwt plugin config is non-conformant. **Consequence:** OIDC conformance test suites fail. Standard RP libraries (`openid-client`, `next-auth`, Spring Security) that default to RS256 for ID Token verification reject tokens signed with an unadvertised algorithm. ## 2. No per-client `id_token_signed_response_alg` The `oauthClient` schema has no `id_token_signed_response_alg` field. [OIDC Dynamic Client Registration 1.0](https://openid.net/specs/openid-connect-registration-1_0.html) defines this as standard client metadata with a default of RS256. **Consequence:** Clients cannot negotiate their preferred signing algorithm during registration. Providers that support multiple algorithms (e.g., RS256 + EdDSA) have no mechanism to sign per-client. ## 3. Discovery only supports advertising a single algorithm `oidcServerMetadata` reads one `alg` from `keyPairConfig` and wraps it in an array. There is no way to advertise multiple supported algorithms (e.g., `["RS256", "EdDSA"]`), even though OIDC Discovery defines `id_token_signing_alg_values_supported` as an array for this purpose. ## 4. Custom `sign` requires `remoteUrl`, which disables built-in JWKS The jwt plugin enforces: `jwt.sign` → requires `jwks.remoteUrl` → disables the built-in `/jwks` endpoint (returns 404). This means adding a secondary signing algorithm (e.g., a post-quantum algorithm alongside EdDSA) forces the developer to take over the entire key management and JWKS serving stack, even for the standard algorithm that the built-in path already handles. **Consequence:** There is no way to extend the signing capabilities incrementally. Any custom algorithm requires replacing the entire JWT infrastructure. ## Spec references - [OIDC Discovery 1.0 — Provider Metadata](https://openid.net/specs/openid-connect-discovery-1_0.html#ProviderMetadata) — RS256 MUST requirement - [OIDC Dynamic Client Registration 1.0](https://openid.net/specs/openid-connect-registration-1_0.html) — `id_token_signed_response_alg` definition, RS256 default - [OIDC Core 1.0 §3.1.3.7](https://openid.net/specs/openid-connect-core-1_0.html#IDTokenValidation) — "The alg value SHOULD be the default of RS256 or the algorithm sent by the Client in the `id_token_signed_response_alg` parameter during Registration"
GiteaMirror added the identity label 2026-04-17 19:49:40 -05:00
GiteaMirror commented 2026-04-17 19:49:44 -05:00
Author
Owner
Copy Link

@dvanmali commented on GitHub (Mar 3, 2026):

  1. Custom sign requires remoteUrl, which disables built-in JWKS

jwt.sign and jwks.remoteUrl were designed for remote signing for key management systems like Google KMS and Amazon EKS, hence /jwks endpoint is disabled since public key logic should be hosted on a CDN or other custom endpoint. PR #4748 related to JWT custom verification could solve custom verification but it appears closed now.

<!-- gh-comment-id:3992737416 --> @dvanmali commented on GitHub (Mar 3, 2026): > 4. Custom sign requires remoteUrl, which disables built-in JWKS `jwt.sign` and `jwks.remoteUrl` were designed for remote signing for key management systems like Google KMS and Amazon EKS, hence `/jwks` endpoint is disabled since public key logic should be hosted on a CDN or other custom endpoint. PR #4748 related to JWT custom verification could solve custom verification but it appears closed now.
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 identity
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#28382
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?