Fix flaky test (#4672)

This issue was due to the fact that the timeout is not always sufficient
to have enough character for truncation + a race between synthetic
timeout and process kill

codex-rs/core/tests/suite/tools.rs

@@ -414,12 +414,15 @@ async fn shell_timeout_includes_timeout_prefix_and_metadata() -> Result<()> {
         .find(|item| item.get("call_id").and_then(Value::as_str) == Some(call_id))
         .expect("timeout output present");
 
-    let output_json: Value = serde_json::from_str(
+    let output_str = timeout_item
-        timeout_item
         .get("output")
         .and_then(Value::as_str)
-            .expect("timeout output string"),
+        .expect("timeout output string");
-    )?;
+
+    // The exec path can report a timeout in two ways depending on timing:
+    // 1) Structured JSON with exit_code 124 and a timeout prefix (preferred), or
+    // 2) A plain error string if the child is observed as killed by a signal first.
+    if let Ok(output_json) = serde_json::from_str::<Value>(output_str) {
         assert_eq!(
             output_json["metadata"]["exit_code"].as_i64(),
             Some(124),
@@ -441,10 +444,17 @@ async fn shell_timeout_includes_timeout_prefix_and_metadata() -> Result<()> {
             duration_ms >= timeout_ms,
             "expected duration >= configured timeout, got {duration_ms} (timeout {timeout_ms})"
         );
+    } else {
+        // Fallback: accept the signal classification path to deflake the test.
         assert!(
-        stdout.contains("[... omitted"),
+            output_str.contains("execution error"),
-        "expected truncated output marker, got {stdout:?}"
+            "unexpected non-JSON output: {output_str:?}"
         );
+        assert!(
+            output_str.contains("Signal(") || output_str.to_lowercase().contains("signal"),
+            "expected signal classification in error output, got {output_str:?}"
+        );
+    }
 
     Ok(())
 }