merge upstream main

backend/internal/repository/account_repo.go

@@ -813,7 +813,7 @@ func (r *accountRepository) SetAntigravityQuotaScopeLimit(ctx context.Context, i
 	client := clientFromContext(ctx, r.client)
 	result, err := client.ExecContext(
 		ctx,
-		`UPDATE accounts SET 
+		`UPDATE accounts SET
 			extra = jsonb_set(
 				jsonb_set(COALESCE(extra, '{}'::jsonb), '{antigravity_quota_scopes}'::text[], COALESCE(extra->'antigravity_quota_scopes', '{}'::jsonb), true),
 				ARRAY['antigravity_quota_scopes', $1]::text[],

backend/internal/service/antigravity_gateway_service.go

@@ -1952,7 +1952,11 @@ func sleepAntigravityBackoffWithContext(ctx context.Context, attempt int) bool {
 
 func antigravityUseScopeRateLimit() bool {
 	v := strings.ToLower(strings.TrimSpace(os.Getenv(antigravityScopeRateLimitEnv)))
-	return v == "1" || v == "true" || v == "yes" || v == "on"
+	// 默认开启按配额域限流，只有明确设置为禁用值时才关闭
+	if v == "0" || v == "false" || v == "no" || v == "off" {
+		return false
+	}
+	return true
 }
 
 func antigravityHasAccountSwitch(ctx context.Context) bool {

backend/internal/service/openai_gateway_service.go

@@ -845,6 +845,12 @@ func (s *OpenAIGatewayService) Forward(ctx context.Context, c *gin.Context, acco
 				bodyModified = true
 			}
 		}
+
+		// Remove prompt_cache_retention (not supported by upstream OpenAI API)
+		if _, has := reqBody["prompt_cache_retention"]; has {
+			delete(reqBody, "prompt_cache_retention")
+			bodyModified = true
+		}
 	}
 
 	// Re-serialize body only if modified