diff --git a/CONTRIBUTING.en.md b/CONTRIBUTING.en.md
index baf5eae..212cbb7 100644
--- a/CONTRIBUTING.en.md
+++ b/CONTRIBUTING.en.md
@@ -86,7 +86,7 @@ Manually build WebUI to `static/admin/`:
 go test ./...
 
 # End-to-end live tests (real accounts)
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 ```
 
 ## Project Structure
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index c75d450..3bbd5b3 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -86,7 +86,7 @@ docker-compose -f docker-compose.dev.yml up
 go test ./...
 
 # 端到端全链路测试（真实账号）
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 ```
 
 ## 项目结构
diff --git a/DEPLOY.en.md b/DEPLOY.en.md
index 8a62c98..046df83 100644
--- a/DEPLOY.en.md
+++ b/DEPLOY.en.md
@@ -518,7 +518,7 @@ curl http://127.0.0.1:5001/v1/chat/completions \
 Run the full live testsuite before release (real account tests):
 
 ```bash
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 ```
 
 With custom flags:
diff --git a/DEPLOY.md b/DEPLOY.md
index e5b0630..4a1efcb 100644
--- a/DEPLOY.md
+++ b/DEPLOY.md
@@ -518,7 +518,7 @@ curl http://127.0.0.1:5001/v1/chat/completions \
 建议在发布前执行完整的端到端测试集（使用真实账号）：
 
 ```bash
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 ```
 
 可自定义参数：
diff --git a/README.MD b/README.MD
index 261e34a..af24e5b 100644
--- a/README.MD
+++ b/README.MD
@@ -350,8 +350,10 @@ ds2api/
 │       ├── components/      # AccountManager / ApiTester / BatchImport / VercelSync / Login / LandingPage
 │       └── locales/         # 中英文语言包（zh.json / en.json）
 ├── scripts/
-│   ├── build-webui.sh       # WebUI 手动构建脚本
-│   └── testsuite/           # 测试集运行脚本
+│   └── build-webui.sh       # WebUI 手动构建脚本
+├── tests/
+│   ├── compat/              # 兼容性测试夹具与期望输出
+│   └── scripts/             # 统一测试脚本入口（unit/e2e）
 ├── static/admin/            # WebUI 构建产物（不提交到 Git）
 ├── .github/
 │   ├── workflows/           # GitHub Actions（Release 自动构建）
@@ -379,11 +381,11 @@ ds2api/
 ## 测试
 
 ```bash
-# 单元测试
-go test ./...
+# 单元测试（Go + Node）
+./tests/scripts/run-unit-all.sh
 
 # 一键端到端全链路测试（真实账号，生成完整请求/响应日志）
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 
 # 或自定义参数
 go run ./cmd/ds2api-tests \
diff --git a/README.en.md b/README.en.md
index 5d2f326..07a5ac8 100644
--- a/README.en.md
+++ b/README.en.md
@@ -350,8 +350,10 @@ ds2api/
 │       ├── components/      # AccountManager / ApiTester / BatchImport / VercelSync / Login / LandingPage
 │       └── locales/         # Language packs (zh.json / en.json)
 ├── scripts/
-│   ├── build-webui.sh       # Manual WebUI build script
-│   └── testsuite/           # Testsuite runner scripts
+│   └── build-webui.sh       # Manual WebUI build script
+├── tests/
+│   ├── compat/              # Compatibility fixtures and expected outputs
+│   └── scripts/             # Unified test script entrypoints (unit/e2e)
 ├── static/admin/            # WebUI build output (not committed to Git)
 ├── .github/
 │   ├── workflows/           # GitHub Actions (Release artifact automation)
@@ -379,11 +381,11 @@ ds2api/
 ## Testing
 
 ```bash
-# Unit tests
-go test ./...
+# Unit tests (Go + Node)
+./tests/scripts/run-unit-all.sh
 
 # One-command live end-to-end tests (real accounts, full request/response logs)
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 
 # Or with custom flags
 go run ./cmd/ds2api-tests \
diff --git a/TESTING.md b/TESTING.md
index ce349ec..f8e532a 100644
--- a/TESTING.md
+++ b/TESTING.md
@@ -8,8 +8,10 @@ DS2API 提供两个层级的测试：
 
 | 层级 | 命令 | 说明 |
 | --- | --- | --- |
-| 单元测试 | `go test ./...` | 不需要真实账号 |
-| 端到端测试 | `./scripts/testsuite/run-live.sh` | 使用真实账号执行全链路测试 |
+| 单元测试（Go） | `./tests/scripts/run-unit-go.sh` | 不需要真实账号 |
+| 单元测试（Node） | `./tests/scripts/run-unit-node.sh` | 不需要真实账号 |
+| 单元测试（全部） | `./tests/scripts/run-unit-all.sh` | 不需要真实账号 |
+| 端到端测试 | `./tests/scripts/run-live.sh` | 使用真实账号执行全链路测试 |
 
 端到端测试集会录制完整的请求/响应日志，用于故障排查。
 
@@ -20,17 +22,19 @@ DS2API 提供两个层级的测试：
 ### 单元测试 | Unit Tests
 
 ```bash
-go test ./...
+./tests/scripts/run-unit-all.sh
 ```
 
 ```bash
-node --test api/helpers/stream-tool-sieve.test.js api/chat-stream.test.js api/compat/js_compat_test.js
+# 或按语言拆分执行
+./tests/scripts/run-unit-go.sh
+./tests/scripts/run-unit-node.sh
 ```
 
 ### 端到端测试 | End-to-End Tests
 
 ```bash
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 ```
 
 **默认行为**：
@@ -179,7 +183,7 @@ go run ./cmd/ds2api-tests \
 
 ```bash
 # 确保 config.json 存在且包含有效测试账号
-./scripts/testsuite/run-live.sh
+./tests/scripts/run-live.sh
 exit_code=$?
 if [ $exit_code -ne 0 ]; then
   echo "Tests failed! Check artifacts for details."
diff --git a/internal/format/openai/render.go b/internal/format/openai/render.go
index 02c536e..248c7ca 100644
--- a/internal/format/openai/render.go
+++ b/internal/format/openai/render.go
@@ -47,27 +47,46 @@ func BuildResponseObject(responseID, model, finalPrompt, finalThinking, finalTex
 	// produced a standalone structured payload. This prevents accidental
 	// empty output_text on normal prose that merely contains tool_call-like text.
 	detected := util.ParseStandaloneToolCalls(finalText, toolNames)
+	toolCallsFromThinking := false
+	if len(detected) == 0 && strings.TrimSpace(finalThinking) != "" {
+		detected = util.ParseStandaloneToolCalls(finalThinking, toolNames)
+		toolCallsFromThinking = len(detected) > 0
+	}
 	exposedOutputText := finalText
 	output := make([]any, 0, 2)
 	if len(detected) > 0 {
-		exposedOutputText = ""
+		if !toolCallsFromThinking || strings.TrimSpace(finalText) != "" {
+			exposedOutputText = ""
+		} else {
+			exposedOutputText = finalThinking
+		}
+		if strings.TrimSpace(finalThinking) != "" {
+			output = append(output, map[string]any{
+				"type": "reasoning",
+				"text": finalThinking,
+			})
+		}
 		output = append(output, map[string]any{
 			"type":       "tool_calls",
 			"tool_calls": util.FormatOpenAIToolCalls(detected),
 		})
 	} else {
-		content := []any{
-			map[string]any{
-				"type": "output_text",
-				"text": finalText,
-			},
-		}
+		content := make([]any, 0, 2)
 		if finalThinking != "" {
 			content = append([]any{map[string]any{
 				"type": "reasoning",
 				"text": finalThinking,
 			}}, content...)
 		}
+		if strings.TrimSpace(finalText) != "" {
+			content = append(content, map[string]any{
+				"type": "output_text",
+				"text": finalText,
+			})
+		}
+		if strings.TrimSpace(finalText) == "" && strings.TrimSpace(finalThinking) != "" {
+			exposedOutputText = finalThinking
+		}
 		output = append(output, map[string]any{
 			"type":    "message",
 			"id":      "msg_" + strings.ReplaceAll(uuid.NewString(), "-", ""),
diff --git a/internal/format/openai/render_test.go b/internal/format/openai/render_test.go
index 50589e0..8fc13e2 100644
--- a/internal/format/openai/render_test.go
+++ b/internal/format/openai/render_test.go
@@ -87,3 +87,60 @@ func TestBuildResponseObjectKeepsOutputTextForMixedProse(t *testing.T) {
 		t.Fatalf("expected output type message, got %#v", first["type"])
 	}
 }
+
+func TestBuildResponseObjectReasoningOnlyFallsBackToOutputText(t *testing.T) {
+	obj := BuildResponseObject(
+		"resp_test",
+		"gpt-4o",
+		"prompt",
+		"internal thinking content",
+		"",
+		nil,
+	)
+
+	outputText, _ := obj["output_text"].(string)
+	if outputText == "" {
+		t.Fatalf("expected output_text fallback from reasoning when final text is empty")
+	}
+
+	output, _ := obj["output"].([]any)
+	if len(output) != 1 {
+		t.Fatalf("expected one output item, got %#v", obj["output"])
+	}
+	first, _ := output[0].(map[string]any)
+	if first["type"] != "message" {
+		t.Fatalf("expected output type message, got %#v", first["type"])
+	}
+	content, _ := first["content"].([]any)
+	if len(content) == 0 {
+		t.Fatalf("expected reasoning content, got %#v", first["content"])
+	}
+	block0, _ := content[0].(map[string]any)
+	if block0["type"] != "reasoning" {
+		t.Fatalf("expected first content block reasoning, got %#v", block0["type"])
+	}
+}
+
+func TestBuildResponseObjectDetectsToolCallFromThinkingChannel(t *testing.T) {
+	obj := BuildResponseObject(
+		"resp_test",
+		"gpt-4o",
+		"prompt",
+		`{"tool_calls":[{"name":"search","input":{"q":"from-thinking"}}]}`,
+		"",
+		[]string{"search"},
+	)
+
+	output, _ := obj["output"].([]any)
+	if len(output) != 2 {
+		t.Fatalf("expected reasoning + tool_calls outputs, got %#v", obj["output"])
+	}
+	first, _ := output[0].(map[string]any)
+	if first["type"] != "reasoning" {
+		t.Fatalf("expected first output reasoning, got %#v", first["type"])
+	}
+	second, _ := output[1].(map[string]any)
+	if second["type"] != "tool_calls" {
+		t.Fatalf("expected second output tool_calls, got %#v", second["type"])
+	}
+}
diff --git a/scripts/testsuite/run-live.sh b/tests/scripts/run-live.sh
similarity index 100%
rename from scripts/testsuite/run-live.sh
rename to tests/scripts/run-live.sh
diff --git a/tests/scripts/run-unit-all.sh b/tests/scripts/run-unit-all.sh
new file mode 100755
index 0000000..59b202c
--- /dev/null
+++ b/tests/scripts/run-unit-all.sh
@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "$0")/../.." && pwd)"
+cd "$ROOT_DIR"
+
+./tests/scripts/run-unit-go.sh
+./tests/scripts/run-unit-node.sh
diff --git a/tests/scripts/run-unit-go.sh b/tests/scripts/run-unit-go.sh
new file mode 100755
index 0000000..38a11b8
--- /dev/null
+++ b/tests/scripts/run-unit-go.sh
@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "$0")/../.." && pwd)"
+cd "$ROOT_DIR"
+
+go test ./... "$@"
diff --git a/tests/scripts/run-unit-node.sh b/tests/scripts/run-unit-node.sh
new file mode 100755
index 0000000..95f11e0
--- /dev/null
+++ b/tests/scripts/run-unit-node.sh
@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+set -euo pipefail
+
+ROOT_DIR="$(cd "$(dirname "$0")/../.." && pwd)"
+cd "$ROOT_DIR"
+
+node --test api/helpers/stream-tool-sieve.test.js api/chat-stream.test.js api/compat/js_compat_test.js "$@"