{"version":"2.1.305","anchor":"retry-after-propagated-on-4295xx-for-an-http-client","canonical_anchor":"retry-after-propagated-on-4295xx-for-an-http-client","heading":"Rate-limit and server errors now carry the server's requested wait","tier":"internal","area":"Internals","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.305\/e\/retry-after-propagated-on-4295xx-for-an-http-client","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.305","markdown":"### Rate-limit and server errors now carry the server's requested wait\n\nWhen a web service answers with a 429 or 5xx error, Claude Code now keeps the server's requested wait time so a retry can honour it\n\n**What**\n\nOne of the parts of Claude Code that talks to a web service raises an error when the service answers with code 429 (too many requests) or a 5xx code (a server-side error). Previously the error held only the status code. Now, for 429, for 503, and for any answer that includes a `retry-after` header, the error also carries how long the server asked Claude Code to wait.\n\n**Why**\n\nCode that retries the request can now wait as long as the server asked, rather than retrying too soon.\n\n- Area: Internals\n- Tier: Under the hood\n- Useful: 1\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}