From 7bb919a0a0d0198ddd84ee0621e53196e85a54d1 Mon Sep 17 00:00:00 2001 From: Andy Dixon Date: Thu, 26 Feb 2026 11:48:40 +0000 Subject: [PATCH] docs: recommend bounded retry/backoff for token exchange failures Update integration guidance to avoid tight retry loops during transient token endpoint failures by using bounded backoff strategies. --- docs/INTEGRATION_GUIDE.md | 1 + 1 file changed, 1 insertion(+) diff --git a/docs/INTEGRATION_GUIDE.md b/docs/INTEGRATION_GUIDE.md index ec89778..63e293a 100644 --- a/docs/INTEGRATION_GUIDE.md +++ b/docs/INTEGRATION_GUIDE.md @@ -140,6 +140,7 @@ try { - Avoid printing tokens in production pages. - Expect `429 Too Many Requests` from provider control endpoints under abuse protection; implement backoff/retry instead of tight loops. - Apply the same backoff strategy for provider `userinfo` calls that return `429 Too Many Requests`. +- For token exchange and refresh, apply bounded retry/backoff on transient provider failures instead of immediate repeated retries. - If you call provider WebAuthn login challenge/verify endpoints directly, use `POST` only. - Configure OAuth client redirect URIs as HTTPS in production; reserve HTTP for localhost loopback testing only. - Rotate client secrets for confidential clients.