From cec8960549998e868130fd5e3070f9d666b1fd6f Mon Sep 17 00:00:00 2001
From: daniel <daniel@speakeasy.com>
Date: Mon, 12 Oct 2026 16:58:28 +0100
Subject: [PATCH] fix(chatgpt): retry transient WebSocket response failures

---
 Cargo.lock                              |  4 +--
 Cargo.toml                              |  2 +-
 src/provider/chatgpt/websocket_tests.rs | 33 +++++++++++++++++++++++++
 3 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index fb9d4e0b..9e6cb7fe 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -327,9 +327,9 @@ dependencies = [
 
 [[package]]
 name = "agentkit-provider-openai"
-version = "0.10.12"
+version = "0.10.13"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "59949238158745f1c53d3f31057dda3b7a643581f067b91b993075e6b3b09511"
+checksum = "aea3cc16b94aa7bc6e55e403c8224d27eb6838af31626fd9404ab888e99f4ca6"
 dependencies = [
  "agentkit-adapter-completions",
  "agentkit-core",
diff --git a/Cargo.toml b/Cargo.toml
index 1a89cd6e..22b30754 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -49,7 +49,7 @@ agentkit-loop = { version = "=0.10.13", features = ["otel"] }
 agentkit-mcp = "=0.10.7"
 agentkit-http = "=0.10.7"
 agentkit-plugins = "=0.10.7"
-agentkit-provider-openai = "=0.10.12"
+agentkit-provider-openai = "=0.10.13"
 agentkit-provider-openrouter = "=0.10.8"
 agentkit-task-manager = "=0.10.7"
 agentkit-tool-compose = { version = "=0.10.12", default-features = false, features = ["runlet"] }
diff --git a/src/provider/chatgpt/websocket_tests.rs b/src/provider/chatgpt/websocket_tests.rs
index d568d500..b2a8b614 100644
--- a/src/provider/chatgpt/websocket_tests.rs
+++ b/src/provider/chatgpt/websocket_tests.rs
@@ -502,6 +502,39 @@ async fn subscription_retries_reach_the_session_retry_observer() {
     assert!(retries.load(std::sync::atomic::Ordering::SeqCst) > 0);
 }
 
+#[tokio::test]
+async fn subscription_retries_correlated_transient_response_failure() {
+    let (endpoint, peer) = server(|listener| {
+        let mut failed = tungstenite::accept(accept(&listener)).unwrap();
+        receive(&mut failed, false);
+        failed
+            .send(Message::Text(response("failed", "")[0].to_string().into()))
+            .unwrap();
+        failed
+            .send(Message::Text(
+                json!({
+                    "type": "response.failed",
+                    "response": {
+                        "id": "failed",
+                        "error": {"type": "server_error", "code": "server_error", "status": 503}
+                    }
+                })
+                .to_string()
+                .into(),
+            ))
+            .unwrap();
+        // Keep the failed socket open: recovery must follow the terminal event,
+        // not a transport disconnect. The replacement sends the full request.
+        let mut fresh = tungstenite::accept(accept(&listener)).unwrap();
+        receive(&mut fresh, false);
+        success(&mut fresh, "retried", "first answer");
+    });
+    let mut session = session_with_retries(&endpoint, 1).await;
+    let mut turn = begin(&mut session, false).await;
+    finished(&mut turn, "retried", "first answer").await;
+    peer.join().unwrap();
+}
+
 #[tokio::test]
 async fn subscription_auto_426_fallback_stays_http_for_future_turns() {
     let (endpoint, peer) = server(|listener| {