feat: add DisableCooling configuration to manage quota cooldown behavior

- Updated the Execute methods in various executors (GeminiCLIExecutor, GeminiExecutor, IFlowExecutor, OpenAICompatExecutor, QwenExecutor) to return a response and error as named return values for improved clarity.
- Enhanced error handling by deferring failure tracking in usage reporters, ensuring that failures are reported correctly.
- Improved response body handling by ensuring proper closure and error logging for HTTP responses across all executors.
- Added failure tracking and reporting in the usage reporter to capture unsuccessful requests.
- Updated the usage logging structure to include a 'Failed' field for better tracking of request outcomes.
- Adjusted the logic in the RequestStatistics and Record methods to accommodate the new failure tracking mechanism.

README.md

@@ -82,6 +82,8 @@ A web-based management center for CLIProxyAPI.
 
 Set `remote-management.disable-control-panel` to `true` if you prefer to host the management UI elsewhere; the server will skip downloading `management.html` and `/management.html` will return 404.
 
+You can set the `MANAGEMENT_STATIC_PATH` environment variable to choose the directory where `management.html` is stored.
+
 ### Authentication
 
 You can authenticate for Gemini, OpenAI, Claude, Qwen, and/or iFlow. All can coexist in the same `auth-dir` and will be load balanced.
@@ -464,6 +466,7 @@ An S3-compatible object storage service can host configuration and authenticatio
 
 | Variable                 | Required | Default                        | Description                                                                                                              |
 |--------------------------|----------|--------------------------------|--------------------------------------------------------------------------------------------------------------------------|
+| `MANAGEMENT_PASSWORD`    | Yes      |                                | Password for the management web UI (required when remote management is enabled).                                        |
 | `OBJECTSTORE_ENDPOINT`   | Yes      |                                | Object storage endpoint. Include `http://` or `https://` to force the protocol (omitted scheme → HTTPS).                |
 | `OBJECTSTORE_BUCKET`     | Yes      |                                | Bucket that stores `config/config.yaml` and `auths/*.json`.                                                             |
 | `OBJECTSTORE_ACCESS_KEY` | Yes      |                                | Access key ID for the object storage account.                                                                           |
@@ -529,21 +532,6 @@ And you can always use Gemini CLI with `CODE_ASSIST_ENDPOINT` set to `http://127
 
 The `auth-dir` parameter specifies where authentication tokens are stored. When you run the login command, the application will create JSON files in this directory containing the authentication tokens for your Google accounts. Multiple accounts can be used for load balancing.
 
-### Request Authentication Providers
-
-Configure inbound authentication through the `auth.providers` section. The built-in `config-api-key` provider works with inline keys:
-
-```
-auth:
-  providers:
-    - name: default
-      type: config-api-key
-      api-keys:
-        - your-api-key-1
-```
-
-Clients should send requests with an `Authorization: Bearer your-api-key-1` header (or `X-Goog-Api-Key`, `X-Api-Key`, or `?key=` as before). The legacy top-level `api-keys` array is still accepted and automatically synced to the default provider for backwards compatibility.
-
 ### Official Generative Language API
 
 The `generative-language-api-key` parameter allows you to define a list of API keys that can be used to authenticate requests to the official Generative Language API.
@@ -809,6 +797,10 @@ Those projects are based on CLIProxyAPI:
 
 Native macOS menu bar app to use your Claude Code & ChatGPT subscriptions with AI coding tools - no API keys needed
 
+### [Subtitle Translator](https://github.com/VjayC/SRT-Subtitle-Translator-Validator)
+
+Browser-based tool to translate SRT subtitles using your Gemini subscription via CLIProxyAPI with automatic validation/error correction - no API keys needed
+
 > [!NOTE]  
 > If you developed a project based on CLIProxyAPI, please open a PR to add it to this list.
 

README_CN.md

@@ -96,6 +96,8 @@ CLIProxyAPI 的基于 Web 的管理中心。
 
 如果希望自行托管管理页面，可在配置中将 `remote-management.disable-control-panel` 设为 `true`，服务器将停止下载 `management.html`，并让 `/management.html` 返回 404。
 
+可以通过设置环境变量 `MANAGEMENT_STATIC_PATH` 来指定 `management.html` 的存储目录。
+
 ### 身份验证
 
 您可以分别为 Gemini、OpenAI、Claude、Qwen 和 iFlow 进行身份验证，它们可同时存在于同一个 `auth-dir` 中并参与负载均衡。
@@ -436,7 +438,7 @@ openai-compatibility:
 
 | 变量                      | 必需 | 默认值    | 描述                                                 |
 |-------------------------|----|--------|----------------------------------------------------|
-| `MANAGEMENT_PASSWORD`   | 是  |        | 控制面板密码                                             |
+| `MANAGEMENT_PASSWORD`   | 是  |        | 管理面板密码                                             |
 | `GITSTORE_GIT_URL`      | 是  |        | 要使用的 Git 仓库的 HTTPS URL。                            |
 | `GITSTORE_LOCAL_PATH`   | 否  | 当前工作目录 | 将克隆 Git 仓库的本地路径。在 Docker 内部，此路径默认为 `/CLIProxyAPI`。 |
 | `GITSTORE_GIT_USERNAME` | 否  |        | 用于 Git 身份验证的用户名。                                   |
@@ -477,6 +479,7 @@ openai-compatibility:
 
 | 变量                     | 是否必填 | 默认值             | 说明                                                                                                                     |
 |--------------------------|----------|--------------------|--------------------------------------------------------------------------------------------------------------------------|
+| `MANAGEMENT_PASSWORD`    | 是       |                    | 管理面板密码（启用远程管理时必需）。                                                                             |
 | `OBJECTSTORE_ENDPOINT`   | 是       |                    | 对象存储访问端点。可带 `http://` 或 `https://` 前缀指定协议（省略则默认 HTTPS）。                                      |
 | `OBJECTSTORE_BUCKET`     | 是       |                    | 用于存放 `config/config.yaml` 与 `auths/*.json` 的 Bucket 名称。                                                        |
 | `OBJECTSTORE_ACCESS_KEY` | 是       |                    | 对象存储账号的访问密钥 ID。                                                                                              |
@@ -537,21 +540,6 @@ openai-compatibility:
 
 `auth-dir` 参数指定身份验证令牌的存储位置。当您运行登录命令时，应用程序将在此目录中创建包含 Google 账户身份验证令牌的 JSON 文件。多个账户可用于轮询。
 
-### 请求鉴权提供方
-
-通过 `auth.providers` 配置接入请求鉴权。内置的 `config-api-key` 提供方支持内联密钥：
-
-```
-auth:
-  providers:
-    - name: default
-      type: config-api-key
-      api-keys:
-        - your-api-key-1
-```
-
-调用时可在 `Authorization` 标头中携带密钥（或继续使用 `X-Goog-Api-Key`、`X-Api-Key`、查询参数 `key`）。为了兼容旧版本，顶层的 `api-keys` 字段仍然可用，并会自动同步到默认的 `config-api-key` 提供方。
-
 ### 官方生成式语言 API
 
 `generative-language-api-key` 参数允许您定义可用于验证对官方 AIStudio Gemini API 请求的 API 密钥列表。
@@ -819,6 +807,10 @@ docker run --rm -p 8317:8317 -v /path/to/your/config.yaml:/CLIProxyAPI/config.ya
 
 一个原生 macOS 菜单栏应用，让您可以使用 Claude Code & ChatGPT 订阅服务和 AI 编程工具，无需 API 密钥。
 
+### [Subtitle Translator](https://github.com/VjayC/SRT-Subtitle-Translator-Validator)
+
+一款基于浏览器的 SRT 字幕翻译工具，可通过 CLI 代理 API 使用您的 Gemini 订阅。内置自动验证与错误修正功能，无需 API 密钥。
+
 > [!NOTE]  
 > 如果你开发了基于 CLIProxyAPI 的项目，请提交一个 PR（拉取请求）将其添加到此列表中。
 

cmd/server/main.go

@@ -20,12 +20,14 @@ import (
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/cmd"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/logging"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/managementasset"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/misc"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/store"
 	_ "github.com/router-for-me/CLIProxyAPI/v6/internal/translator"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/usage"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	sdkAuth "github.com/router-for-me/CLIProxyAPI/v6/sdk/auth"
+	coreauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
 	log "github.com/sirupsen/logrus"
 )
 
@@ -147,6 +149,7 @@ func main() {
 		}
 		return "", false
 	}
+	writableBase := util.WritablePath()
 	if value, ok := lookupEnv("PGSTORE_DSN", "pgstore_dsn"); ok {
 		usePostgresStore = true
 		pgStoreDSN = value
@@ -158,6 +161,13 @@ func main() {
 		if value, ok := lookupEnv("PGSTORE_LOCAL_PATH", "pgstore_local_path"); ok {
 			pgStoreLocalPath = value
 		}
+		if pgStoreLocalPath == "" {
+			if writableBase != "" {
+				pgStoreLocalPath = writableBase
+			} else {
+				pgStoreLocalPath = wd
+			}
+		}
 		useGitStore = false
 	}
 	if value, ok := lookupEnv("GITSTORE_GIT_URL", "gitstore_git_url"); ok {
@@ -229,11 +239,14 @@ func main() {
 			log.Infof("postgres-backed token store enabled, workspace path: %s", pgStoreInst.WorkDir())
 		}
 	} else if useObjectStore {
-		objectStoreRoot := objectStoreLocalPath
-		if objectStoreRoot == "" {
-			objectStoreRoot = wd
+		if objectStoreLocalPath == "" {
+			if writableBase != "" {
+				objectStoreLocalPath = writableBase
+			} else {
+				objectStoreLocalPath = wd
+			}
 		}
-		objectStoreRoot = filepath.Join(objectStoreRoot, "objectstore")
+		objectStoreRoot := filepath.Join(objectStoreLocalPath, "objectstore")
 		resolvedEndpoint := strings.TrimSpace(objectStoreEndpoint)
 		useSSL := true
 		if strings.Contains(resolvedEndpoint, "://") {
@@ -289,7 +302,11 @@ func main() {
 		}
 	} else if useGitStore {
 		if gitStoreLocalPath == "" {
-			gitStoreLocalPath = wd
+			if writableBase != "" {
+				gitStoreLocalPath = writableBase
+			} else {
+				gitStoreLocalPath = wd
+			}
 		}
 		gitStoreRoot = filepath.Join(gitStoreLocalPath, "gitstore")
 		authDir := filepath.Join(gitStoreRoot, "auths")
@@ -361,6 +378,7 @@ func main() {
 		}
 	}
 	usage.SetStatisticsEnabled(cfg.UsageStatisticsEnabled)
+	coreauth.SetQuotaCooldownDisabled(cfg.DisableCooling)
 
 	if err = logging.ConfigureLogOutput(cfg.LoggingToFile); err != nil {
 		log.Fatalf("failed to configure log output: %v", err)
@@ -376,6 +394,7 @@ func main() {
 	} else {
 		cfg.AuthDir = resolvedAuthDir
 	}
+	managementasset.SetCurrentConfig(cfg)
 
 	// Create login options to be used in authentication flows.
 	options := &cmd.LoginOptions{
@@ -419,6 +438,7 @@ func main() {
 			return
 		}
 		// Start the main proxy service
+		managementasset.StartAutoUpdater(context.Background(), configFilePath)
 		cmd.StartService(cfg, configFilePath, password)
 	}
 }

internal/api/handlers/management/config_basic.go

@@ -1,13 +1,126 @@
 package management
 
 import (
+	"io"
+	"net/http"
+	"os"
+	"path/filepath"
+
 	"github.com/gin-gonic/gin"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
+	"gopkg.in/yaml.v3"
 )
 
 func (h *Handler) GetConfig(c *gin.Context) {
 	c.JSON(200, h.cfg)
 }
 
+func (h *Handler) GetConfigYAML(c *gin.Context) {
+	data, err := os.ReadFile(h.configFilePath)
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "read_failed", "message": err.Error()})
+		return
+	}
+	var node yaml.Node
+	if err := yaml.Unmarshal(data, &node); err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "parse_failed", "message": err.Error()})
+		return
+	}
+	c.Header("Content-Type", "application/yaml; charset=utf-8")
+	c.Header("Vary", "format, Accept")
+	enc := yaml.NewEncoder(c.Writer)
+	enc.SetIndent(2)
+	_ = enc.Encode(&node)
+	_ = enc.Close()
+}
+
+func WriteConfig(path string, data []byte) error {
+	f, err := os.OpenFile(path, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0644)
+	if err != nil {
+		return err
+	}
+	if _, err := f.Write(data); err != nil {
+		f.Close()
+		return err
+	}
+	if err := f.Sync(); err != nil {
+		f.Close()
+		return err
+	}
+	return f.Close()
+}
+
+func (h *Handler) PutConfigYAML(c *gin.Context) {
+	body, err := io.ReadAll(c.Request.Body)
+	if err != nil {
+		c.JSON(http.StatusBadRequest, gin.H{"error": "invalid_yaml", "message": "cannot read request body"})
+		return
+	}
+	var cfg config.Config
+	if err := yaml.Unmarshal(body, &cfg); err != nil {
+		c.JSON(http.StatusBadRequest, gin.H{"error": "invalid_yaml", "message": err.Error()})
+		return
+	}
+	// Validate config using LoadConfigOptional with optional=false to enforce parsing
+	tmpDir := filepath.Dir(h.configFilePath)
+	tmpFile, err := os.CreateTemp(tmpDir, "config-validate-*.yaml")
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "write_failed", "message": err.Error()})
+		return
+	}
+	tempFile := tmpFile.Name()
+	if _, err := tmpFile.Write(body); err != nil {
+		tmpFile.Close()
+		os.Remove(tempFile)
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "write_failed", "message": err.Error()})
+		return
+	}
+	if err := tmpFile.Close(); err != nil {
+		os.Remove(tempFile)
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "write_failed", "message": err.Error()})
+		return
+	}
+	defer os.Remove(tempFile)
+	_, err = config.LoadConfigOptional(tempFile, false)
+	if err != nil {
+		c.JSON(http.StatusUnprocessableEntity, gin.H{"error": "invalid_config", "message": err.Error()})
+		return
+	}
+	h.mu.Lock()
+	defer h.mu.Unlock()
+	if WriteConfig(h.configFilePath, body) != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "write_failed", "message": "failed to write config"})
+		return
+	}
+	// Reload into handler to keep memory in sync
+	newCfg, err := config.LoadConfig(h.configFilePath)
+	if err != nil {
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "reload_failed", "message": err.Error()})
+		return
+	}
+	h.cfg = newCfg
+	c.JSON(http.StatusOK, gin.H{"ok": true, "changed": []string{"config"}})
+}
+
+// GetConfigFile returns the raw config.yaml file bytes without re-encoding.
+// It preserves comments and original formatting/styles.
+func (h *Handler) GetConfigFile(c *gin.Context) {
+	data, err := os.ReadFile(h.configFilePath)
+	if err != nil {
+		if os.IsNotExist(err) {
+			c.JSON(http.StatusNotFound, gin.H{"error": "not_found", "message": "config file not found"})
+			return
+		}
+		c.JSON(http.StatusInternalServerError, gin.H{"error": "read_failed", "message": err.Error()})
+		return
+	}
+	c.Header("Content-Type", "application/yaml; charset=utf-8")
+	c.Header("Cache-Control", "no-store")
+	c.Header("X-Content-Type-Options", "nosniff")
+	// Write raw bytes as-is
+	_, _ = c.Writer.Write(data)
+}
+
 // Debug
 func (h *Handler) GetDebug(c *gin.Context) { c.JSON(200, gin.H{"debug": h.cfg.Debug}) }
 func (h *Handler) PutDebug(c *gin.Context) { h.updateBoolField(c, func(v bool) { h.cfg.Debug = v }) }

internal/api/handlers/management/logs.go

@@ -13,6 +13,7 @@ import (
 	"time"
 
 	"github.com/gin-gonic/gin"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 )
 
 const (
@@ -145,6 +146,9 @@ func (h *Handler) logDirectory() string {
 	if h.logDir != "" {
 		return h.logDir
 	}
+	if base := util.WritablePath(); base != "" {
+		return filepath.Join(base, "logs")
+	}
 	if h.configFilePath != "" {
 		dir := filepath.Dir(h.configFilePath)
 		if dir != "" && dir != "." {

internal/api/handlers/management/usage.go

@@ -13,5 +13,8 @@ func (h *Handler) GetUsageStatistics(c *gin.Context) {
 	if h != nil && h.usageStats != nil {
 		snapshot = h.usageStats.Snapshot()
 	}
-	c.JSON(http.StatusOK, gin.H{"usage": snapshot})
+	c.JSON(http.StatusOK, gin.H{
+		"usage":           snapshot,
+		"failed_requests": snapshot.FailureCount,
+	})
 }

internal/api/server.go

@@ -52,7 +52,11 @@ type serverOptionConfig struct {
 type ServerOption func(*serverOptionConfig)
 
 func defaultRequestLoggerFactory(cfg *config.Config, configPath string) logging.RequestLogger {
-	return logging.NewFileRequestLogger(cfg.RequestLog, "logs", filepath.Dir(configPath))
+	configDir := filepath.Dir(configPath)
+	if base := util.WritablePath(); base != "" {
+		return logging.NewFileRequestLogger(cfg.RequestLog, filepath.Join(base, "logs"), configDir)
+	}
+	return logging.NewFileRequestLogger(cfg.RequestLog, "logs", configDir)
 }
 
 // WithMiddleware appends additional Gin middleware during server construction.
@@ -228,12 +232,18 @@ func NewServer(cfg *config.Config, authManager *auth.Manager, accessManager *sdk
 	// Save initial YAML snapshot
 	s.oldConfigYaml, _ = yaml.Marshal(cfg)
 	s.applyAccessConfig(nil, cfg)
+	managementasset.SetCurrentConfig(cfg)
+	auth.SetQuotaCooldownDisabled(cfg.DisableCooling)
 	// Initialize management handler
 	s.mgmt = managementHandlers.NewHandler(cfg, configFilePath, authManager)
 	if optionState.localPassword != "" {
 		s.mgmt.SetLocalPassword(optionState.localPassword)
 	}
-	s.mgmt.SetLogDirectory(filepath.Join(s.currentPath, "logs"))
+	logDir := filepath.Join(s.currentPath, "logs")
+	if base := util.WritablePath(); base != "" {
+		logDir = filepath.Join(base, "logs")
+	}
+	s.mgmt.SetLogDirectory(logDir)
 	s.localPassword = optionState.localPassword
 
 	// Setup routes
@@ -376,6 +386,8 @@ func (s *Server) registerManagementRoutes() {
 	{
 		mgmt.GET("/usage", s.mgmt.GetUsageStatistics)
 		mgmt.GET("/config", s.mgmt.GetConfig)
+		mgmt.PUT("/config.yaml", s.mgmt.PutConfigYAML)
+		mgmt.GET("/config.yaml", s.mgmt.GetConfigFile)
 
 		mgmt.GET("/debug", s.mgmt.GetDebug)
 		mgmt.PUT("/debug", s.mgmt.PutDebug)
@@ -705,6 +717,15 @@ func (s *Server) UpdateClients(cfg *config.Config) {
 		}
 	}
 
+	if oldCfg == nil || oldCfg.DisableCooling != cfg.DisableCooling {
+		auth.SetQuotaCooldownDisabled(cfg.DisableCooling)
+		if oldCfg != nil {
+			log.Debugf("disable_cooling updated from %t to %t", oldCfg.DisableCooling, cfg.DisableCooling)
+		} else {
+			log.Debugf("disable_cooling toggled to %t", cfg.DisableCooling)
+		}
+	}
+
 	// Update log level dynamically when debug flag changes
 	if oldCfg == nil || oldCfg.Debug != cfg.Debug {
 		util.SetLogLevel(cfg)
@@ -749,6 +770,7 @@ func (s *Server) UpdateClients(cfg *config.Config) {
 
 	s.applyAccessConfig(oldCfg, cfg)
 	s.cfg = cfg
+	managementasset.SetCurrentConfig(cfg)
 	// Save YAML snapshot for next comparison
 	s.oldConfigYaml, _ = yaml.Marshal(cfg)
 	s.handlers.UpdateClients(&cfg.SDKConfig)

internal/config/config.go

@@ -34,6 +34,9 @@ type Config struct {
 	// UsageStatisticsEnabled toggles in-memory usage aggregation; when false, usage data is discarded.
 	UsageStatisticsEnabled bool `yaml:"usage-statistics-enabled" json:"usage-statistics-enabled"`
 
+	// DisableCooling disables quota cooldown scheduling when true.
+	DisableCooling bool `yaml:"disable-cooling" json:"disable-cooling"`
+
 	// QuotaExceeded defines the behavior when a quota is exceeded.
 	QuotaExceeded QuotaExceeded `yaml:"quota-exceeded" json:"quota-exceeded"`
 
@@ -183,6 +186,7 @@ func LoadConfigOptional(configFile string, optional bool) (*Config, error) {
 	// Set defaults before unmarshal so that absent keys keep defaults.
 	cfg.LoggingToFile = false
 	cfg.UsageStatisticsEnabled = false
+	cfg.DisableCooling = false
 	if err = yaml.Unmarshal(data, &cfg); err != nil {
 		if optional {
 			// In cloud deploy mode, if YAML parsing fails, return empty config instead of error.

internal/logging/global_logger.go

@@ -10,6 +10,7 @@ import (
 	"sync"
 
 	"github.com/gin-gonic/gin"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	log "github.com/sirupsen/logrus"
 	"gopkg.in/natefinch/lumberjack.v2"
 )
@@ -72,7 +73,10 @@ func ConfigureLogOutput(loggingToFile bool) error {
 	defer writerMu.Unlock()
 
 	if loggingToFile {
-		const logDir = "logs"
+		logDir := "logs"
+		if base := util.WritablePath(); base != "" {
+			logDir = filepath.Join(base, "logs")
+		}
 		if err := os.MkdirAll(logDir, 0o755); err != nil {
 			return fmt.Errorf("logging: failed to create log directory: %w", err)
 		}

internal/logging/request_logger.go

@@ -16,6 +16,7 @@ import (
 	"time"
 
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 )
 
 // RequestLogger defines the interface for logging HTTP requests and responses.
@@ -328,9 +329,19 @@ func (l *FileRequestLogger) formatLogContent(url, method string, headers map[str
 	// Request info
 	content.WriteString(l.formatRequestInfo(url, method, headers, body))
 
-	content.WriteString("=== API REQUEST ===\n")
-	content.Write(apiRequest)
-	content.WriteString("\n\n")
+	if len(apiRequest) > 0 {
+		if bytes.HasPrefix(apiRequest, []byte("=== API REQUEST")) {
+			content.Write(apiRequest)
+			if !bytes.HasSuffix(apiRequest, []byte("\n")) {
+				content.WriteString("\n")
+			}
+		} else {
+			content.WriteString("=== API REQUEST ===\n")
+			content.Write(apiRequest)
+			content.WriteString("\n")
+		}
+		content.WriteString("\n")
+	}
 
 	for i := 0; i < len(apiResponseErrors); i++ {
 		content.WriteString("=== API ERROR RESPONSE ===\n")
@@ -339,9 +350,19 @@ func (l *FileRequestLogger) formatLogContent(url, method string, headers map[str
 		content.WriteString("\n\n")
 	}
 
-	content.WriteString("=== API RESPONSE ===\n")
-	content.Write(apiResponse)
-	content.WriteString("\n\n")
+	if len(apiResponse) > 0 {
+		if bytes.HasPrefix(apiResponse, []byte("=== API RESPONSE")) {
+			content.Write(apiResponse)
+			if !bytes.HasSuffix(apiResponse, []byte("\n")) {
+				content.WriteString("\n")
+			}
+		} else {
+			content.WriteString("=== API RESPONSE ===\n")
+			content.Write(apiResponse)
+			content.WriteString("\n")
+		}
+		content.WriteString("\n")
+	}
 
 	// Response section
 	content.WriteString("=== RESPONSE ===\n")
@@ -465,7 +486,8 @@ func (l *FileRequestLogger) formatRequestInfo(url, method string, headers map[st
 	content.WriteString("=== HEADERS ===\n")
 	for key, values := range headers {
 		for _, value := range values {
-			content.WriteString(fmt.Sprintf("%s: %s\n", key, value))
+			masked := util.MaskSensitiveHeaderValue(key, value)
+			content.WriteString(fmt.Sprintf("%s: %s\n", key, masked))
 		}
 	}
 	content.WriteString("\n")

internal/managementasset/updater.go

@@ -13,8 +13,10 @@ import (
 	"path/filepath"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	sdkconfig "github.com/router-for-me/CLIProxyAPI/v6/sdk/config"
 	log "github.com/sirupsen/logrus"
@@ -33,8 +35,83 @@ const ManagementFileName = managementAssetName
 var (
 	lastUpdateCheckMu   sync.Mutex
 	lastUpdateCheckTime time.Time
+
+	currentConfigPtr    atomic.Pointer[config.Config]
+	disableControlPanel atomic.Bool
+	schedulerOnce       sync.Once
+	schedulerConfigPath atomic.Value
 )
 
+// SetCurrentConfig stores the latest configuration snapshot for management asset decisions.
+func SetCurrentConfig(cfg *config.Config) {
+	if cfg == nil {
+		currentConfigPtr.Store(nil)
+		return
+	}
+
+	prevDisabled := disableControlPanel.Load()
+	currentConfigPtr.Store(cfg)
+	disableControlPanel.Store(cfg.RemoteManagement.DisableControlPanel)
+
+	if prevDisabled && !cfg.RemoteManagement.DisableControlPanel {
+		lastUpdateCheckMu.Lock()
+		lastUpdateCheckTime = time.Time{}
+		lastUpdateCheckMu.Unlock()
+	}
+}
+
+// StartAutoUpdater launches a background goroutine that periodically ensures the management asset is up to date.
+// It respects the disable-control-panel flag on every iteration and supports hot-reloaded configurations.
+func StartAutoUpdater(ctx context.Context, configFilePath string) {
+	configFilePath = strings.TrimSpace(configFilePath)
+	if configFilePath == "" {
+		log.Debug("management asset auto-updater skipped: empty config path")
+		return
+	}
+
+	schedulerConfigPath.Store(configFilePath)
+
+	schedulerOnce.Do(func() {
+		go runAutoUpdater(ctx)
+	})
+}
+
+func runAutoUpdater(ctx context.Context) {
+	if ctx == nil {
+		ctx = context.Background()
+	}
+
+	ticker := time.NewTicker(updateCheckInterval)
+	defer ticker.Stop()
+
+	runOnce := func() {
+		cfg := currentConfigPtr.Load()
+		if cfg == nil {
+			log.Debug("management asset auto-updater skipped: config not yet available")
+			return
+		}
+		if disableControlPanel.Load() {
+			log.Debug("management asset auto-updater skipped: control panel disabled")
+			return
+		}
+
+		configPath, _ := schedulerConfigPath.Load().(string)
+		staticDir := StaticDir(configPath)
+		EnsureLatestManagementHTML(ctx, staticDir, cfg.ProxyURL)
+	}
+
+	runOnce()
+
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case <-ticker.C:
+			runOnce()
+		}
+	}
+}
+
 func newHTTPClient(proxyURL string) *http.Client {
 	client := &http.Client{Timeout: 15 * time.Second}
 
@@ -56,6 +133,18 @@ type releaseResponse struct {
 
 // StaticDir resolves the directory that stores the management control panel asset.
 func StaticDir(configFilePath string) string {
+	if override := strings.TrimSpace(os.Getenv("MANAGEMENT_STATIC_PATH")); override != "" {
+		cleaned := filepath.Clean(override)
+		if strings.EqualFold(filepath.Base(cleaned), managementAssetName) {
+			return filepath.Dir(cleaned)
+		}
+		return cleaned
+	}
+
+	if writable := util.WritablePath(); writable != "" {
+		return filepath.Join(writable, "static")
+	}
+
 	configFilePath = strings.TrimSpace(configFilePath)
 	if configFilePath == "" {
 		return ""
@@ -74,6 +163,14 @@ func StaticDir(configFilePath string) string {
 
 // FilePath resolves the absolute path to the management control panel asset.
 func FilePath(configFilePath string) string {
+	if override := strings.TrimSpace(os.Getenv("MANAGEMENT_STATIC_PATH")); override != "" {
+		cleaned := filepath.Clean(override)
+		if strings.EqualFold(filepath.Base(cleaned), managementAssetName) {
+			return cleaned
+		}
+		return filepath.Join(cleaned, ManagementFileName)
+	}
+
 	dir := StaticDir(configFilePath)
 	if dir == "" {
 		return ""
@@ -89,6 +186,11 @@ func EnsureLatestManagementHTML(ctx context.Context, staticDir string, proxyURL
 		ctx = context.Background()
 	}
 
+	if disableControlPanel.Load() {
+		log.Debug("management asset sync skipped: control panel disabled by configuration")
+		return
+	}
+
 	staticDir = strings.TrimSpace(staticDir)
 	if staticDir == "" {
 		log.Debug("management asset sync skipped: empty static directory")

internal/runtime/executor/cache_helpers.go

@@ -0,0 +1,10 @@
+package executor
+
+import "time"
+
+type codexCache struct {
+	ID     string
+	Expire time.Time
+}
+
+var codexCacheMap = map[string]codexCache{}

internal/runtime/executor/claude_executor.go

@@ -36,13 +36,14 @@ func (e *ClaudeExecutor) Identifier() string { return "claude" }
 
 func (e *ClaudeExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
 
-func (e *ClaudeExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *ClaudeExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	apiKey, baseURL := claudeCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://api.anthropic.com"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("claude")
 	// Use streaming translation to preserve function calling, except for claude.
@@ -54,42 +55,63 @@ func (e *ClaudeExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 	}
 
 	url := fmt.Sprintf("%s/v1/messages?beta=true", baseURL)
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	applyClaudeHeaders(httpReq, apiKey, false)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	defer func() {
-		if errClose := resp.Body.Close(); errClose != nil {
+		if errClose := httpResp.Body.Close(); errClose != nil {
 			log.Errorf("response body close error: %v", errClose)
 		}
 	}()
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
-	reader := io.Reader(resp.Body)
+	reader := io.Reader(httpResp.Body)
 	var decoder *zstd.Decoder
-	if hasZSTDEcoding(resp.Header.Get("Content-Encoding")) {
-		decoder, err = zstd.NewReader(resp.Body)
+	if hasZSTDEcoding(httpResp.Header.Get("Content-Encoding")) {
+		decoder, err = zstd.NewReader(httpResp.Body)
 		if err != nil {
-			return cliproxyexecutor.Response{}, fmt.Errorf("failed to initialize zstd decoder: %w", err)
+			recordAPIResponseError(ctx, e.cfg, err)
+			return resp, fmt.Errorf("failed to initialize zstd decoder: %w", err)
 		}
 		reader = decoder
 		defer decoder.Close()
 	}
 	data, err := io.ReadAll(reader)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)
 	if stream {
@@ -104,49 +126,77 @@ func (e *ClaudeExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 	}
 	var param any
 	out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, data, &param)
-	return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+	resp = cliproxyexecutor.Response{Payload: []byte(out)}
+	return resp, nil
 }
 
-func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	apiKey, baseURL := claudeCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://api.anthropic.com"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("claude")
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
 	body, _ = sjson.SetRawBytes(body, "system", []byte(misc.ClaudeCodeInstructions))
 
 	url := fmt.Sprintf("%s/v1/messages?beta=true", baseURL)
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
 	}
 	applyClaudeHeaders(httpReq, apiKey, true)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("response body close error: %v", errClose)
+		}
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return nil, err
 	}
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("response body close error: %v", errClose)
+			}
+		}()
 
 		// If from == to (Claude → Claude), directly forward the SSE stream without translation
 		if from == to {
-			scanner := bufio.NewScanner(resp.Body)
+			scanner := bufio.NewScanner(httpResp.Body)
 			buf := make([]byte, 20_971_520)
 			scanner.Buffer(buf, 20_971_520)
 			for scanner.Scan() {
@@ -161,14 +211,16 @@ func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 				cloned[len(line)] = '\n'
 				out <- cliproxyexecutor.StreamChunk{Payload: cloned}
 			}
-			if err = scanner.Err(); err != nil {
-				out <- cliproxyexecutor.StreamChunk{Err: err}
+			if errScan := scanner.Err(); errScan != nil {
+				recordAPIResponseError(ctx, e.cfg, errScan)
+				reporter.publishFailure(ctx)
+				out <- cliproxyexecutor.StreamChunk{Err: errScan}
 			}
 			return
 		}
 
 		// For other formats, use translation
-		scanner := bufio.NewScanner(resp.Body)
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -183,11 +235,13 @@ func (e *ClaudeExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
 			}
 		}
-		if err = scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
-	return out, nil
+	return stream, nil
 }
 
 func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
@@ -208,16 +262,33 @@ func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	}
 
 	url := fmt.Sprintf("%s/v1/messages/count_tokens?beta=true", baseURL)
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return cliproxyexecutor.Response{}, err
 	}
 	applyClaudeHeaders(httpReq, apiKey, false)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
 	resp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return cliproxyexecutor.Response{}, err
 	}
 	defer func() {
@@ -225,6 +296,7 @@ func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 			log.Errorf("response body close error: %v", errClose)
 		}
 	}()
+	recordAPIResponseMetadata(ctx, e.cfg, resp.StatusCode, resp.Header.Clone())
 	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
 		b, _ := io.ReadAll(resp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
@@ -235,6 +307,7 @@ func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	if hasZSTDEcoding(resp.Header.Get("Content-Encoding")) {
 		decoder, err = zstd.NewReader(resp.Body)
 		if err != nil {
+			recordAPIResponseError(ctx, e.cfg, err)
 			return cliproxyexecutor.Response{}, fmt.Errorf("failed to initialize zstd decoder: %w", err)
 		}
 		reader = decoder
@@ -242,6 +315,7 @@ func (e *ClaudeExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	}
 	data, err := io.ReadAll(reader)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return cliproxyexecutor.Response{}, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)

internal/runtime/executor/codex_executor.go

@@ -39,13 +39,14 @@ func (e *CodexExecutor) Identifier() string { return "codex" }
 
 func (e *CodexExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
 
-func (e *CodexExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *CodexExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	apiKey, baseURL := codexCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://chatgpt.com/backend-api/codex"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("codex")
@@ -78,29 +79,75 @@ func (e *CodexExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, re
 	body, _ = sjson.SetBytes(body, "stream", true)
 	body, _ = sjson.DeleteBytes(body, "previous_response_id")
 
+	additionalHeaders := make(map[string]string)
+	if from == "claude" {
+		userIDResult := gjson.GetBytes(req.Payload, "metadata.user_id")
+		if userIDResult.Exists() {
+			var cache codexCache
+			var hasKey bool
+			key := fmt.Sprintf("%s-%s", req.Model, userIDResult.String())
+			if cache, hasKey = codexCacheMap[key]; !hasKey || cache.Expire.Before(time.Now()) {
+				cache = codexCache{
+					ID:     uuid.New().String(),
+					Expire: time.Now().Add(1 * time.Hour),
+				}
+				codexCacheMap[key] = cache
+			}
+			additionalHeaders["Conversation_id"] = cache.ID
+			additionalHeaders["Session_id"] = cache.ID
+			body, _ = sjson.SetBytes(body, "prompt_cache_key", cache.ID)
+		}
+	}
+
 	url := strings.TrimSuffix(baseURL, "/") + "/responses"
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	applyCodexHeaders(httpReq, auth, apiKey)
-
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	for k, v := range additionalHeaders {
+		httpReq.Header.Set(k, v)
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
-	defer func() { _ = resp.Body.Close() }()
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	defer func() {
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("codex executor: close response body error: %v", errClose)
+		}
+	}()
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
-	data, err := io.ReadAll(resp.Body)
+	data, err := io.ReadAll(httpResp.Body)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)
 
@@ -121,18 +168,21 @@ func (e *CodexExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, re
 
 		var param any
 		out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, line, &param)
-		return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+		resp = cliproxyexecutor.Response{Payload: []byte(out)}
+		return resp, nil
 	}
-	return cliproxyexecutor.Response{}, statusErr{code: 408, msg: "stream error: stream disconnected before completion: stream closed before response.completed"}
+	err = statusErr{code: 408, msg: "stream error: stream disconnected before completion: stream closed before response.completed"}
+	return resp, err
 }
 
-func (e *CodexExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *CodexExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	apiKey, baseURL := codexCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://chatgpt.com/backend-api/codex"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("codex")
@@ -164,31 +214,84 @@ func (e *CodexExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
 
 	body, _ = sjson.DeleteBytes(body, "previous_response_id")
 
+	additionalHeaders := make(map[string]string)
+	if from == "claude" {
+		userIDResult := gjson.GetBytes(req.Payload, "metadata.user_id")
+		if userIDResult.Exists() {
+			var cache codexCache
+			var hasKey bool
+			key := fmt.Sprintf("%s-%s", req.Model, userIDResult.String())
+			if cache, hasKey = codexCacheMap[key]; !hasKey || cache.Expire.Before(time.Now()) {
+				cache = codexCache{
+					ID:     uuid.New().String(),
+					Expire: time.Now().Add(1 * time.Hour),
+				}
+				codexCacheMap[key] = cache
+			}
+			additionalHeaders["Conversation_id"] = cache.ID
+			additionalHeaders["Session_id"] = cache.ID
+			body, _ = sjson.SetBytes(body, "prompt_cache_key", cache.ID)
+		}
+	}
+
 	url := strings.TrimSuffix(baseURL, "/") + "/responses"
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
 	}
 	applyCodexHeaders(httpReq, auth, apiKey)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	for k, v := range additionalHeaders {
+		httpReq.Header.Set(k, v)
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
-		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		data, readErr := io.ReadAll(httpResp.Body)
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("codex executor: close response body error: %v", errClose)
+		}
+		if readErr != nil {
+			recordAPIResponseError(ctx, e.cfg, readErr)
+			return nil, readErr
+		}
+		appendAPIResponseChunk(ctx, e.cfg, data)
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(data))
+		err = statusErr{code: httpResp.StatusCode, msg: string(data)}
+		return nil, err
 	}
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
-		scanner := bufio.NewScanner(resp.Body)
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("codex executor: close response body error: %v", errClose)
+			}
+		}()
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -210,11 +313,13 @@ func (e *CodexExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
 			}
 		}
-		if err = scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
-	return out, nil
+	return stream, nil
 }
 
 func (e *CodexExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {

internal/runtime/executor/gemini_cli_executor.go

@@ -51,12 +51,13 @@ func (e *GeminiCLIExecutor) Identifier() string { return "gemini-cli" }
 
 func (e *GeminiCLIExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
 
-func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	tokenSource, baseTokenData, err := prepareGeminiCLITokenSource(ctx, e.cfg, auth)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("gemini-cli")
@@ -83,6 +84,11 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
 	httpClient := newHTTPClient(ctx, e.cfg, auth, 0)
 	respCtx := context.WithValue(ctx, "alt", opts.Alt)
 
+	var authID, authLabel, authType, authValue string
+	authID = auth.ID
+	authLabel = auth.Label
+	authType, authValue = auth.AccountInfo()
+
 	var lastStatus int
 	var lastBody []byte
 
@@ -99,7 +105,8 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
 
 		tok, errTok := tokenSource.Token()
 		if errTok != nil {
-			return cliproxyexecutor.Response{}, errTok
+			err = errTok
+			return resp, err
 		}
 		updateGeminiCLITokenMetadata(auth, baseTokenData, tok)
 
@@ -108,48 +115,81 @@ func (e *GeminiCLIExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth
 			url = url + fmt.Sprintf("?$alt=%s", opts.Alt)
 		}
 
-		recordAPIRequest(ctx, e.cfg, payload)
 		reqHTTP, errReq := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(payload))
 		if errReq != nil {
-			return cliproxyexecutor.Response{}, errReq
+			err = errReq
+			return resp, err
 		}
 		reqHTTP.Header.Set("Content-Type", "application/json")
 		reqHTTP.Header.Set("Authorization", "Bearer "+tok.AccessToken)
 		applyGeminiCLIHeaders(reqHTTP)
 		reqHTTP.Header.Set("Accept", "application/json")
+		recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+			URL:       url,
+			Method:    http.MethodPost,
+			Headers:   reqHTTP.Header.Clone(),
+			Body:      payload,
+			Provider:  e.Identifier(),
+			AuthID:    authID,
+			AuthLabel: authLabel,
+			AuthType:  authType,
+			AuthValue: authValue,
+		})
 
-		resp, errDo := httpClient.Do(reqHTTP)
+		httpResp, errDo := httpClient.Do(reqHTTP)
 		if errDo != nil {
-			return cliproxyexecutor.Response{}, errDo
+			recordAPIResponseError(ctx, e.cfg, errDo)
+			err = errDo
+			return resp, err
+		}
+
+		data, errRead := io.ReadAll(httpResp.Body)
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("gemini cli executor: close response body error: %v", errClose)
+		}
+		recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+		if errRead != nil {
+			recordAPIResponseError(ctx, e.cfg, errRead)
+			err = errRead
+			return resp, err
 		}
-		data, _ := io.ReadAll(resp.Body)
-		_ = resp.Body.Close()
 		appendAPIResponseChunk(ctx, e.cfg, data)
-		if resp.StatusCode >= 200 && resp.StatusCode < 300 {
+		if httpResp.StatusCode >= 200 && httpResp.StatusCode < 300 {
 			reporter.publish(ctx, parseGeminiCLIUsage(data))
 			var param any
 			out := sdktranslator.TranslateNonStream(respCtx, to, from, attemptModel, bytes.Clone(opts.OriginalRequest), payload, data, &param)
-			return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+			resp = cliproxyexecutor.Response{Payload: []byte(out)}
+			return resp, nil
 		}
-		lastStatus = resp.StatusCode
-		lastBody = data
-		if resp.StatusCode != 429 {
-			break
+
+		lastStatus = httpResp.StatusCode
+		lastBody = append([]byte(nil), data...)
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(data))
+		if httpResp.StatusCode == 429 {
+			continue
 		}
+
+		err = statusErr{code: httpResp.StatusCode, msg: string(data)}
+		return resp, err
 	}
 
 	if len(lastBody) > 0 {
 		appendAPIResponseChunk(ctx, e.cfg, lastBody)
 	}
-	return cliproxyexecutor.Response{}, statusErr{code: lastStatus, msg: string(lastBody)}
+	if lastStatus == 0 {
+		lastStatus = 429
+	}
+	err = statusErr{code: lastStatus, msg: string(lastBody)}
+	return resp, err
 }
 
-func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	tokenSource, baseTokenData, err := prepareGeminiCLITokenSource(ctx, e.cfg, auth)
 	if err != nil {
 		return nil, err
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("gemini-cli")
@@ -170,6 +210,11 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 	httpClient := newHTTPClient(ctx, e.cfg, auth, 0)
 	respCtx := context.WithValue(ctx, "alt", opts.Alt)
 
+	var authID, authLabel, authType, authValue string
+	authID = auth.ID
+	authLabel = auth.Label
+	authType, authValue = auth.AccountInfo()
+
 	var lastStatus int
 	var lastBody []byte
 
@@ -181,7 +226,8 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 
 		tok, errTok := tokenSource.Token()
 		if errTok != nil {
-			return nil, errTok
+			err = errTok
+			return nil, err
 		}
 		updateGeminiCLITokenMetadata(auth, baseTokenData, tok)
 
@@ -192,37 +238,64 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 			url = url + fmt.Sprintf("?$alt=%s", opts.Alt)
 		}
 
-		recordAPIRequest(ctx, e.cfg, payload)
 		reqHTTP, errReq := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(payload))
 		if errReq != nil {
-			return nil, errReq
+			err = errReq
+			return nil, err
 		}
 		reqHTTP.Header.Set("Content-Type", "application/json")
 		reqHTTP.Header.Set("Authorization", "Bearer "+tok.AccessToken)
 		applyGeminiCLIHeaders(reqHTTP)
 		reqHTTP.Header.Set("Accept", "text/event-stream")
+		recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+			URL:       url,
+			Method:    http.MethodPost,
+			Headers:   reqHTTP.Header.Clone(),
+			Body:      payload,
+			Provider:  e.Identifier(),
+			AuthID:    authID,
+			AuthLabel: authLabel,
+			AuthType:  authType,
+			AuthValue: authValue,
+		})
 
-		resp, errDo := httpClient.Do(reqHTTP)
+		httpResp, errDo := httpClient.Do(reqHTTP)
 		if errDo != nil {
-			return nil, errDo
+			recordAPIResponseError(ctx, e.cfg, errDo)
+			err = errDo
+			return nil, err
 		}
-		if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-			data, _ := io.ReadAll(resp.Body)
-			_ = resp.Body.Close()
+		recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+		if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+			data, errRead := io.ReadAll(httpResp.Body)
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("gemini cli executor: close response body error: %v", errClose)
+			}
+			if errRead != nil {
+				recordAPIResponseError(ctx, e.cfg, errRead)
+				err = errRead
+				return nil, err
+			}
 			appendAPIResponseChunk(ctx, e.cfg, data)
-			lastStatus = resp.StatusCode
-			lastBody = data
-			log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(data))
-			if resp.StatusCode == 429 {
+			lastStatus = httpResp.StatusCode
+			lastBody = append([]byte(nil), data...)
+			log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(data))
+			if httpResp.StatusCode == 429 {
 				continue
 			}
-			return nil, statusErr{code: resp.StatusCode, msg: string(data)}
+			err = statusErr{code: httpResp.StatusCode, msg: string(data)}
+			return nil, err
 		}
 
 		out := make(chan cliproxyexecutor.StreamChunk)
+		stream = out
 		go func(resp *http.Response, reqBody []byte, attempt string) {
 			defer close(out)
-			defer func() { _ = resp.Body.Close() }()
+			defer func() {
+				if errClose := resp.Body.Close(); errClose != nil {
+					log.Errorf("gemini cli executor: close response body error: %v", errClose)
+				}
+			}()
 			if opts.Alt == "" {
 				scanner := bufio.NewScanner(resp.Body)
 				buf := make([]byte, 20_971_520)
@@ -247,6 +320,8 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 					out <- cliproxyexecutor.StreamChunk{Payload: []byte(segments[i])}
 				}
 				if errScan := scanner.Err(); errScan != nil {
+					recordAPIResponseError(ctx, e.cfg, errScan)
+					reporter.publishFailure(ctx)
 					out <- cliproxyexecutor.StreamChunk{Err: errScan}
 				}
 				return
@@ -254,6 +329,8 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 
 			data, errRead := io.ReadAll(resp.Body)
 			if errRead != nil {
+				recordAPIResponseError(ctx, e.cfg, errRead)
+				reporter.publishFailure(ctx)
 				out <- cliproxyexecutor.StreamChunk{Err: errRead}
 				return
 			}
@@ -269,15 +346,19 @@ func (e *GeminiCLIExecutor) ExecuteStream(ctx context.Context, auth *cliproxyaut
 			for i := range segments {
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(segments[i])}
 			}
-		}(resp, append([]byte(nil), payload...), attemptModel)
+		}(httpResp, append([]byte(nil), payload...), attemptModel)
 
-		return out, nil
+		return stream, nil
 	}
 
+	if len(lastBody) > 0 {
+		appendAPIResponseChunk(ctx, e.cfg, lastBody)
+	}
 	if lastStatus == 0 {
 		lastStatus = 429
 	}
-	return nil, statusErr{code: lastStatus, msg: string(lastBody)}
+	err = statusErr{code: lastStatus, msg: string(lastBody)}
+	return nil, err
 }
 
 func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
@@ -297,6 +378,13 @@ func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.
 	httpClient := newHTTPClient(ctx, e.cfg, auth, 0)
 	respCtx := context.WithValue(ctx, "alt", opts.Alt)
 
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+
 	var lastStatus int
 	var lastBody []byte
 
@@ -322,7 +410,6 @@ func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.
 			url = url + fmt.Sprintf("?$alt=%s", opts.Alt)
 		}
 
-		recordAPIRequest(ctx, e.cfg, payload)
 		reqHTTP, errReq := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(payload))
 		if errReq != nil {
 			return cliproxyexecutor.Response{}, errReq
@@ -331,13 +418,30 @@ func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.
 		reqHTTP.Header.Set("Authorization", "Bearer "+tok.AccessToken)
 		applyGeminiCLIHeaders(reqHTTP)
 		reqHTTP.Header.Set("Accept", "application/json")
+		recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+			URL:       url,
+			Method:    http.MethodPost,
+			Headers:   reqHTTP.Header.Clone(),
+			Body:      payload,
+			Provider:  e.Identifier(),
+			AuthID:    authID,
+			AuthLabel: authLabel,
+			AuthType:  authType,
+			AuthValue: authValue,
+		})
 
 		resp, errDo := httpClient.Do(reqHTTP)
 		if errDo != nil {
+			recordAPIResponseError(ctx, e.cfg, errDo)
 			return cliproxyexecutor.Response{}, errDo
 		}
-		data, _ := io.ReadAll(resp.Body)
+		data, errRead := io.ReadAll(resp.Body)
 		_ = resp.Body.Close()
+		recordAPIResponseMetadata(ctx, e.cfg, resp.StatusCode, resp.Header.Clone())
+		if errRead != nil {
+			recordAPIResponseError(ctx, e.cfg, errRead)
+			return cliproxyexecutor.Response{}, errRead
+		}
 		appendAPIResponseChunk(ctx, e.cfg, data)
 		if resp.StatusCode >= 200 && resp.StatusCode < 300 {
 			count := gjson.GetBytes(data, "totalTokens").Int()
@@ -345,16 +449,13 @@ func (e *GeminiCLIExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.
 			return cliproxyexecutor.Response{Payload: []byte(translated)}, nil
 		}
 		lastStatus = resp.StatusCode
-		lastBody = data
+		lastBody = append([]byte(nil), data...)
 		if resp.StatusCode == 429 {
 			continue
 		}
 		break
 	}
 
-	if len(lastBody) > 0 {
-		appendAPIResponseChunk(ctx, e.cfg, lastBody)
-	}
 	if lastStatus == 0 {
 		lastStatus = 429
 	}

internal/runtime/executor/gemini_executor.go

@@ -68,10 +68,11 @@ func (e *GeminiExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) e
 // Returns:
 //   - cliproxyexecutor.Response: The response from the API
 //   - error: An error if the request fails
-func (e *GeminiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *GeminiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	apiKey, bearer := geminiCreds(auth)
 
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	// Official Gemini API via API key or OAuth bearer
 	from := opts.SourceFormat
@@ -96,10 +97,9 @@ func (e *GeminiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 
 	body, _ = sjson.DeleteBytes(body, "session_id")
 
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	httpReq.Header.Set("Content-Type", "application/json")
 	if apiKey != "" {
@@ -107,34 +107,61 @@ func (e *GeminiExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, r
 	} else if bearer != "" {
 		httpReq.Header.Set("Authorization", "Bearer "+bearer)
 	}
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
-	defer func() { _ = resp.Body.Close() }()
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	defer func() {
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("gemini executor: close response body error: %v", errClose)
+		}
+	}()
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
-	data, err := io.ReadAll(resp.Body)
+	data, err := io.ReadAll(httpResp.Body)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)
 	reporter.publish(ctx, parseGeminiUsage(data))
 	var param any
 	out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, data, &param)
-	return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+	resp = cliproxyexecutor.Response{Payload: []byte(out)}
+	return resp, nil
 }
 
-func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	apiKey, bearer := geminiCreds(auth)
 
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("gemini")
@@ -154,7 +181,6 @@ func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 
 	body, _ = sjson.DeleteBytes(body, "session_id")
 
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
@@ -165,24 +191,51 @@ func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 	} else {
 		httpReq.Header.Set("Authorization", "Bearer "+bearer)
 	}
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("gemini executor: close response body error: %v", errClose)
+		}
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return nil, err
 	}
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
-		scanner := bufio.NewScanner(resp.Body)
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("gemini executor: close response body error: %v", errClose)
+			}
+		}()
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -201,11 +254,13 @@ func (e *GeminiExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.A
 		for i := range lines {
 			out <- cliproxyexecutor.StreamChunk{Payload: []byte(lines[i])}
 		}
-		if err = scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
-	return out, nil
+	return stream, nil
 }
 
 func (e *GeminiExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
@@ -224,7 +279,6 @@ func (e *GeminiExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	translatedReq, _ = sjson.DeleteBytes(translatedReq, "generationConfig")
 
 	url := fmt.Sprintf("%s/%s/models/%s:%s", glEndpoint, glAPIVersion, req.Model, "countTokens")
-	recordAPIRequest(ctx, e.cfg, translatedReq)
 
 	requestBody := bytes.NewReader(translatedReq)
 
@@ -238,16 +292,36 @@ func (e *GeminiExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Aut
 	} else {
 		httpReq.Header.Set("Authorization", "Bearer "+bearer)
 	}
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      translatedReq,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
 	resp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return cliproxyexecutor.Response{}, err
 	}
 	defer func() { _ = resp.Body.Close() }()
+	recordAPIResponseMetadata(ctx, e.cfg, resp.StatusCode, resp.Header.Clone())
 
 	data, err := io.ReadAll(resp.Body)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return cliproxyexecutor.Response{}, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)

internal/runtime/executor/iflow_executor.go

@@ -12,6 +12,7 @@ import (
 
 	iflowauth "github.com/router-for-me/CLIProxyAPI/v6/internal/auth/iflow"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	cliproxyauth "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/auth"
 	cliproxyexecutor "github.com/router-for-me/CLIProxyAPI/v6/sdk/cliproxy/executor"
 	sdktranslator "github.com/router-for-me/CLIProxyAPI/v6/sdk/translator"
@@ -40,67 +41,96 @@ func (e *IFlowExecutor) Identifier() string { return "iflow" }
 func (e *IFlowExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
 
 // Execute performs a non-streaming chat completion request.
-func (e *IFlowExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *IFlowExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	apiKey, baseURL := iflowCreds(auth)
 	if strings.TrimSpace(apiKey) == "" {
-		return cliproxyexecutor.Response{}, fmt.Errorf("iflow executor: missing api key")
+		err = fmt.Errorf("iflow executor: missing api key")
+		return resp, err
 	}
 	if baseURL == "" {
 		baseURL = iflowauth.DefaultAPIBaseURL
 	}
 
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 
 	endpoint := strings.TrimSuffix(baseURL, "/") + iflowDefaultEndpoint
-	recordAPIRequest(ctx, e.cfg, body)
 
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewReader(body))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	applyIFlowHeaders(httpReq, apiKey, false)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       endpoint,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
-	defer func() { _ = resp.Body.Close() }()
+	defer func() {
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("iflow executor: close response body error: %v", errClose)
+		}
+	}()
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
 
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("iflow request error: status %d body %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("iflow request error: status %d body %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
 
-	data, err := io.ReadAll(resp.Body)
+	data, err := io.ReadAll(httpResp.Body)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)
 	reporter.publish(ctx, parseOpenAIUsage(data))
 
 	var param any
 	out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, data, &param)
-	return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+	resp = cliproxyexecutor.Response{Payload: []byte(out)}
+	return resp, nil
 }
 
 // ExecuteStream performs a streaming chat completion request.
-func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	apiKey, baseURL := iflowCreds(auth)
 	if strings.TrimSpace(apiKey) == "" {
-		return nil, fmt.Errorf("iflow executor: missing api key")
+		err = fmt.Errorf("iflow executor: missing api key")
+		return nil, err
 	}
 	if baseURL == "" {
 		baseURL = iflowauth.DefaultAPIBaseURL
 	}
 
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
@@ -113,34 +143,60 @@ func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
 	}
 
 	endpoint := strings.TrimSuffix(baseURL, "/") + iflowDefaultEndpoint
-	recordAPIRequest(ctx, e.cfg, body)
 
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, endpoint, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
 	}
 	applyIFlowHeaders(httpReq, apiKey, true)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       endpoint,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
 
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
-		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("iflow streaming error: status %d body %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		data, _ := io.ReadAll(httpResp.Body)
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("iflow executor: close response body error: %v", errClose)
+		}
+		appendAPIResponseChunk(ctx, e.cfg, data)
+		log.Debugf("iflow streaming error: status %d body %s", httpResp.StatusCode, string(data))
+		err = statusErr{code: httpResp.StatusCode, msg: string(data)}
+		return nil, err
 	}
 
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("iflow executor: close response body error: %v", errClose)
+			}
+		}()
 
-		scanner := bufio.NewScanner(resp.Body)
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -155,12 +211,14 @@ func (e *IFlowExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Au
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
 			}
 		}
-		if err := scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
 
-	return out, nil
+	return stream, nil
 }
 
 // CountTokens is not implemented for iFlow.
@@ -176,18 +234,28 @@ func (e *IFlowExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (*
 	}
 
 	refreshToken := ""
+	oldAccessToken := ""
 	if auth.Metadata != nil {
 		if v, ok := auth.Metadata["refresh_token"].(string); ok {
 			refreshToken = strings.TrimSpace(v)
 		}
+		if v, ok := auth.Metadata["access_token"].(string); ok {
+			oldAccessToken = strings.TrimSpace(v)
+		}
 	}
 	if refreshToken == "" {
 		return auth, nil
 	}
 
+	// Log the old access token (masked) before refresh
+	if oldAccessToken != "" {
+		log.Debugf("iflow executor: refreshing access token, old: %s", util.HideAPIKey(oldAccessToken))
+	}
+
 	svc := iflowauth.NewIFlowAuth(e.cfg)
 	tokenData, err := svc.RefreshTokens(ctx, refreshToken)
 	if err != nil {
+		log.Errorf("iflow executor: token refresh failed: %v", err)
 		return nil, err
 	}
 
@@ -205,6 +273,9 @@ func (e *IFlowExecutor) Refresh(ctx context.Context, auth *cliproxyauth.Auth) (*
 	auth.Metadata["type"] = "iflow"
 	auth.Metadata["last_refresh"] = time.Now().Format(time.RFC3339)
 
+	// Log the new access token (masked) after successful refresh
+	log.Debugf("iflow executor: token refresh successful, new: %s", util.HideAPIKey(tokenData.AccessToken))
+
 	if auth.Attributes == nil {
 		auth.Attributes = make(map[string]string)
 	}

internal/runtime/executor/logging_helpers.go

@@ -3,19 +3,144 @@ package executor
 import (
 	"bytes"
 	"context"
+	"fmt"
+	"net/http"
+	"sort"
+	"strings"
+	"time"
 
 	"github.com/gin-gonic/gin"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 )
 
-// recordAPIRequest stores the upstream request payload in Gin context for request logging.
-func recordAPIRequest(ctx context.Context, cfg *config.Config, payload []byte) {
-	if cfg == nil || !cfg.RequestLog || len(payload) == 0 {
+const (
+	apiAttemptsKey = "API_UPSTREAM_ATTEMPTS"
+	apiRequestKey  = "API_REQUEST"
+	apiResponseKey = "API_RESPONSE"
+)
+
+// upstreamRequestLog captures the outbound upstream request details for logging.
+type upstreamRequestLog struct {
+	URL       string
+	Method    string
+	Headers   http.Header
+	Body      []byte
+	Provider  string
+	AuthID    string
+	AuthLabel string
+	AuthType  string
+	AuthValue string
+}
+
+type upstreamAttempt struct {
+	index                int
+	request              string
+	response             *strings.Builder
+	responseIntroWritten bool
+	statusWritten        bool
+	headersWritten       bool
+	bodyStarted          bool
+	bodyHasContent       bool
+	errorWritten         bool
+}
+
+// recordAPIRequest stores the upstream request metadata in Gin context for request logging.
+func recordAPIRequest(ctx context.Context, cfg *config.Config, info upstreamRequestLog) {
+	if cfg == nil || !cfg.RequestLog {
 		return
 	}
-	if ginCtx, ok := ctx.Value("gin").(*gin.Context); ok && ginCtx != nil {
-		ginCtx.Set("API_REQUEST", bytes.Clone(payload))
+	ginCtx := ginContextFrom(ctx)
+	if ginCtx == nil {
+		return
 	}
+
+	attempts := getAttempts(ginCtx)
+	index := len(attempts) + 1
+
+	builder := &strings.Builder{}
+	builder.WriteString(fmt.Sprintf("=== API REQUEST %d ===\n", index))
+	builder.WriteString(fmt.Sprintf("Timestamp: %s\n", time.Now().Format(time.RFC3339Nano)))
+	if info.URL != "" {
+		builder.WriteString(fmt.Sprintf("Upstream URL: %s\n", info.URL))
+	} else {
+		builder.WriteString("Upstream URL: <unknown>\n")
+	}
+	if info.Method != "" {
+		builder.WriteString(fmt.Sprintf("HTTP Method: %s\n", info.Method))
+	}
+	if auth := formatAuthInfo(info); auth != "" {
+		builder.WriteString(fmt.Sprintf("Auth: %s\n", auth))
+	}
+	builder.WriteString("\nHeaders:\n")
+	writeHeaders(builder, info.Headers)
+	builder.WriteString("\nBody:\n")
+	if len(info.Body) > 0 {
+		builder.WriteString(string(bytes.Clone(info.Body)))
+	} else {
+		builder.WriteString("<empty>")
+	}
+	builder.WriteString("\n\n")
+
+	attempt := &upstreamAttempt{
+		index:    index,
+		request:  builder.String(),
+		response: &strings.Builder{},
+	}
+	attempts = append(attempts, attempt)
+	ginCtx.Set(apiAttemptsKey, attempts)
+	updateAggregatedRequest(ginCtx, attempts)
+}
+
+// recordAPIResponseMetadata captures upstream response status/header information for the latest attempt.
+func recordAPIResponseMetadata(ctx context.Context, cfg *config.Config, status int, headers http.Header) {
+	if cfg == nil || !cfg.RequestLog {
+		return
+	}
+	ginCtx := ginContextFrom(ctx)
+	if ginCtx == nil {
+		return
+	}
+	attempts, attempt := ensureAttempt(ginCtx)
+	ensureResponseIntro(attempt)
+
+	if status > 0 && !attempt.statusWritten {
+		attempt.response.WriteString(fmt.Sprintf("Status: %d\n", status))
+		attempt.statusWritten = true
+	}
+	if !attempt.headersWritten {
+		attempt.response.WriteString("Headers:\n")
+		writeHeaders(attempt.response, headers)
+		attempt.headersWritten = true
+		attempt.response.WriteString("\n")
+	}
+
+	updateAggregatedResponse(ginCtx, attempts)
+}
+
+// recordAPIResponseError adds an error entry for the latest attempt when no HTTP response is available.
+func recordAPIResponseError(ctx context.Context, cfg *config.Config, err error) {
+	if cfg == nil || !cfg.RequestLog || err == nil {
+		return
+	}
+	ginCtx := ginContextFrom(ctx)
+	if ginCtx == nil {
+		return
+	}
+	attempts, attempt := ensureAttempt(ginCtx)
+	ensureResponseIntro(attempt)
+
+	if attempt.bodyStarted && !attempt.bodyHasContent {
+		// Ensure body does not stay empty marker if error arrives first.
+		attempt.bodyStarted = false
+	}
+	if attempt.errorWritten {
+		attempt.response.WriteString("\n")
+	}
+	attempt.response.WriteString(fmt.Sprintf("Error: %s\n", err.Error()))
+	attempt.errorWritten = true
+
+	updateAggregatedResponse(ginCtx, attempts)
 }
 
 // appendAPIResponseChunk appends an upstream response chunk to Gin context for request logging.
@@ -27,15 +152,171 @@ func appendAPIResponseChunk(ctx context.Context, cfg *config.Config, chunk []byt
 	if len(data) == 0 {
 		return
 	}
-	if ginCtx, ok := ctx.Value("gin").(*gin.Context); ok && ginCtx != nil {
-		if existing, exists := ginCtx.Get("API_RESPONSE"); exists {
-			if prev, okBytes := existing.([]byte); okBytes {
-				prev = append(prev, data...)
-				prev = append(prev, []byte("\n\n")...)
-				ginCtx.Set("API_RESPONSE", prev)
-				return
-			}
+	ginCtx := ginContextFrom(ctx)
+	if ginCtx == nil {
+		return
+	}
+	attempts, attempt := ensureAttempt(ginCtx)
+	ensureResponseIntro(attempt)
+
+	if !attempt.headersWritten {
+		attempt.response.WriteString("Headers:\n")
+		writeHeaders(attempt.response, nil)
+		attempt.headersWritten = true
+		attempt.response.WriteString("\n")
+	}
+	if !attempt.bodyStarted {
+		attempt.response.WriteString("Body:\n")
+		attempt.bodyStarted = true
+	}
+	if attempt.bodyHasContent {
+		attempt.response.WriteString("\n\n")
+	}
+	attempt.response.WriteString(string(data))
+	attempt.bodyHasContent = true
+
+	updateAggregatedResponse(ginCtx, attempts)
+}
+
+func ginContextFrom(ctx context.Context) *gin.Context {
+	ginCtx, _ := ctx.Value("gin").(*gin.Context)
+	return ginCtx
+}
+
+func getAttempts(ginCtx *gin.Context) []*upstreamAttempt {
+	if ginCtx == nil {
+		return nil
+	}
+	if value, exists := ginCtx.Get(apiAttemptsKey); exists {
+		if attempts, ok := value.([]*upstreamAttempt); ok {
+			return attempts
+		}
+	}
+	return nil
+}
+
+func ensureAttempt(ginCtx *gin.Context) ([]*upstreamAttempt, *upstreamAttempt) {
+	attempts := getAttempts(ginCtx)
+	if len(attempts) == 0 {
+		attempt := &upstreamAttempt{
+			index:    1,
+			request:  "=== API REQUEST 1 ===\n<missing>\n\n",
+			response: &strings.Builder{},
+		}
+		attempts = []*upstreamAttempt{attempt}
+		ginCtx.Set(apiAttemptsKey, attempts)
+		updateAggregatedRequest(ginCtx, attempts)
+	}
+	return attempts, attempts[len(attempts)-1]
+}
+
+func ensureResponseIntro(attempt *upstreamAttempt) {
+	if attempt == nil || attempt.response == nil || attempt.responseIntroWritten {
+		return
+	}
+	attempt.response.WriteString(fmt.Sprintf("=== API RESPONSE %d ===\n", attempt.index))
+	attempt.response.WriteString(fmt.Sprintf("Timestamp: %s\n", time.Now().Format(time.RFC3339Nano)))
+	attempt.response.WriteString("\n")
+	attempt.responseIntroWritten = true
+}
+
+func updateAggregatedRequest(ginCtx *gin.Context, attempts []*upstreamAttempt) {
+	if ginCtx == nil {
+		return
+	}
+	var builder strings.Builder
+	for _, attempt := range attempts {
+		builder.WriteString(attempt.request)
+	}
+	ginCtx.Set(apiRequestKey, []byte(builder.String()))
+}
+
+func updateAggregatedResponse(ginCtx *gin.Context, attempts []*upstreamAttempt) {
+	if ginCtx == nil {
+		return
+	}
+	var builder strings.Builder
+	for idx, attempt := range attempts {
+		if attempt == nil || attempt.response == nil {
+			continue
+		}
+		responseText := attempt.response.String()
+		if responseText == "" {
+			continue
+		}
+		builder.WriteString(responseText)
+		if !strings.HasSuffix(responseText, "\n") {
+			builder.WriteString("\n")
+		}
+		if idx < len(attempts)-1 {
+			builder.WriteString("\n")
+		}
+	}
+	ginCtx.Set(apiResponseKey, []byte(builder.String()))
+}
+
+func writeHeaders(builder *strings.Builder, headers http.Header) {
+	if builder == nil {
+		return
+	}
+	if len(headers) == 0 {
+		builder.WriteString("<none>\n")
+		return
+	}
+	keys := make([]string, 0, len(headers))
+	for key := range headers {
+		keys = append(keys, key)
+	}
+	sort.Strings(keys)
+	for _, key := range keys {
+		values := headers[key]
+		if len(values) == 0 {
+			builder.WriteString(fmt.Sprintf("%s:\n", key))
+			continue
+		}
+		for _, value := range values {
+			masked := util.MaskSensitiveHeaderValue(key, value)
+			builder.WriteString(fmt.Sprintf("%s: %s\n", key, masked))
 		}
-		ginCtx.Set("API_RESPONSE", data)
 	}
 }
+
+func formatAuthInfo(info upstreamRequestLog) string {
+	var parts []string
+	if trimmed := strings.TrimSpace(info.Provider); trimmed != "" {
+		parts = append(parts, fmt.Sprintf("provider=%s", trimmed))
+	}
+	if trimmed := strings.TrimSpace(info.AuthID); trimmed != "" {
+		parts = append(parts, fmt.Sprintf("auth_id=%s", trimmed))
+	}
+	if trimmed := strings.TrimSpace(info.AuthLabel); trimmed != "" {
+		parts = append(parts, fmt.Sprintf("label=%s", trimmed))
+	}
+
+	authType := strings.ToLower(strings.TrimSpace(info.AuthType))
+	authValue := strings.TrimSpace(info.AuthValue)
+	switch authType {
+	case "api_key":
+		if authValue != "" {
+			parts = append(parts, fmt.Sprintf("type=api_key value=%s", util.HideAPIKey(authValue)))
+		} else {
+			parts = append(parts, "type=api_key")
+		}
+	case "oauth":
+		if authValue != "" {
+			parts = append(parts, fmt.Sprintf("type=oauth account=%s", authValue))
+		} else {
+			parts = append(parts, "type=oauth")
+		}
+	default:
+		if authType != "" {
+			if authValue != "" {
+				parts = append(parts, fmt.Sprintf("type=%s value=%s", authType, authValue))
+			} else {
+				parts = append(parts, fmt.Sprintf("type=%s", authType))
+			}
+		}
+	}
+
+	return strings.Join(parts, ", ")
+}

internal/runtime/executor/openai_compat_executor.go

@@ -38,12 +38,15 @@ func (e *OpenAICompatExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.A
 	return nil
 }
 
-func (e *OpenAICompatExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *OpenAICompatExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
+	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
+
 	baseURL, apiKey := e.resolveCredentials(auth)
 	if baseURL == "" {
-		return cliproxyexecutor.Response{}, statusErr{code: http.StatusUnauthorized, msg: "missing provider baseURL"}
+		err = statusErr{code: http.StatusUnauthorized, msg: "missing provider baseURL"}
+		return
 	}
-	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
 
 	// Translate inbound request to OpenAI format
 	from := opts.SourceFormat
@@ -54,47 +57,75 @@ func (e *OpenAICompatExecutor) Execute(ctx context.Context, auth *cliproxyauth.A
 	}
 
 	url := strings.TrimSuffix(baseURL, "/") + "/chat/completions"
-	recordAPIRequest(ctx, e.cfg, translated)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(translated))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	httpReq.Header.Set("Content-Type", "application/json")
 	if apiKey != "" {
 		httpReq.Header.Set("Authorization", "Bearer "+apiKey)
 	}
 	httpReq.Header.Set("User-Agent", "cli-proxy-openai-compat")
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      translated,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
-	defer func() { _ = resp.Body.Close() }()
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	defer func() {
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("openai compat executor: close response body error: %v", errClose)
+		}
+	}()
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
-	body, err := io.ReadAll(resp.Body)
+	body, err := io.ReadAll(httpResp.Body)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, body)
 	reporter.publish(ctx, parseOpenAIUsage(body))
 	// Translate response back to source format when needed
 	var param any
 	out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), translated, body, &param)
-	return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+	resp = cliproxyexecutor.Response{Payload: []byte(out)}
+	return resp, nil
 }
 
-func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
+	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
+
 	baseURL, apiKey := e.resolveCredentials(auth)
 	if baseURL == "" {
-		return nil, statusErr{code: http.StatusUnauthorized, msg: "missing provider baseURL"}
+		err = statusErr{code: http.StatusUnauthorized, msg: "missing provider baseURL"}
+		return nil, err
 	}
-	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
 	translated := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), true)
@@ -103,7 +134,6 @@ func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
 	}
 
 	url := strings.TrimSuffix(baseURL, "/") + "/chat/completions"
-	recordAPIRequest(ctx, e.cfg, translated)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(translated))
 	if err != nil {
 		return nil, err
@@ -115,24 +145,51 @@ func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
 	httpReq.Header.Set("User-Agent", "cli-proxy-openai-compat")
 	httpReq.Header.Set("Accept", "text/event-stream")
 	httpReq.Header.Set("Cache-Control", "no-cache")
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      translated,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("openai compat executor: close response body error: %v", errClose)
+		}
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return nil, err
 	}
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
-		scanner := bufio.NewScanner(resp.Body)
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("openai compat executor: close response body error: %v", errClose)
+			}
+		}()
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -152,11 +209,13 @@ func (e *OpenAICompatExecutor) ExecuteStream(ctx context.Context, auth *cliproxy
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
 			}
 		}
-		if err = scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
-	return out, nil
+	return stream, nil
 }
 
 func (e *OpenAICompatExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {

internal/runtime/executor/qwen_executor.go

@@ -38,56 +38,83 @@ func (e *QwenExecutor) Identifier() string { return "qwen" }
 
 func (e *QwenExecutor) PrepareRequest(_ *http.Request, _ *cliproxyauth.Auth) error { return nil }
 
-func (e *QwenExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {
+func (e *QwenExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (resp cliproxyexecutor.Response, err error) {
 	token, baseURL := qwenCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://portal.qwen.ai/v1"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
 	body := sdktranslator.TranslateRequest(from, to, req.Model, bytes.Clone(req.Payload), false)
 
 	url := strings.TrimSuffix(baseURL, "/") + "/chat/completions"
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		return resp, err
 	}
 	applyQwenHeaders(httpReq, token, false)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
-	defer func() { _ = resp.Body.Close() }()
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		b, _ := io.ReadAll(resp.Body)
+	defer func() {
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("qwen executor: close response body error: %v", errClose)
+		}
+	}()
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return cliproxyexecutor.Response{}, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return resp, err
 	}
-	data, err := io.ReadAll(resp.Body)
+	data, err := io.ReadAll(httpResp.Body)
 	if err != nil {
-		return cliproxyexecutor.Response{}, err
+		recordAPIResponseError(ctx, e.cfg, err)
+		return resp, err
 	}
 	appendAPIResponseChunk(ctx, e.cfg, data)
 	reporter.publish(ctx, parseOpenAIUsage(data))
 	var param any
 	out := sdktranslator.TranslateNonStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, data, &param)
-	return cliproxyexecutor.Response{Payload: []byte(out)}, nil
+	resp = cliproxyexecutor.Response{Payload: []byte(out)}
+	return resp, nil
 }
 
-func (e *QwenExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (<-chan cliproxyexecutor.StreamChunk, error) {
+func (e *QwenExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
 	token, baseURL := qwenCreds(auth)
 
 	if baseURL == "" {
 		baseURL = "https://portal.qwen.ai/v1"
 	}
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
+	defer reporter.trackFailure(ctx, &err)
 
 	from := opts.SourceFormat
 	to := sdktranslator.FromString("openai")
@@ -102,30 +129,56 @@ func (e *QwenExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 	body, _ = sjson.SetBytes(body, "stream_options.include_usage", true)
 
 	url := strings.TrimSuffix(baseURL, "/") + "/chat/completions"
-	recordAPIRequest(ctx, e.cfg, body)
 	httpReq, err := http.NewRequestWithContext(ctx, http.MethodPost, url, bytes.NewReader(body))
 	if err != nil {
 		return nil, err
 	}
 	applyQwenHeaders(httpReq, token, true)
+	var authID, authLabel, authType, authValue string
+	if auth != nil {
+		authID = auth.ID
+		authLabel = auth.Label
+		authType, authValue = auth.AccountInfo()
+	}
+	recordAPIRequest(ctx, e.cfg, upstreamRequestLog{
+		URL:       url,
+		Method:    http.MethodPost,
+		Headers:   httpReq.Header.Clone(),
+		Body:      body,
+		Provider:  e.Identifier(),
+		AuthID:    authID,
+		AuthLabel: authLabel,
+		AuthType:  authType,
+		AuthValue: authValue,
+	})
 
 	httpClient := newProxyAwareHTTPClient(ctx, e.cfg, auth, 0)
-	resp, err := httpClient.Do(httpReq)
+	httpResp, err := httpClient.Do(httpReq)
 	if err != nil {
+		recordAPIResponseError(ctx, e.cfg, err)
 		return nil, err
 	}
-	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		defer func() { _ = resp.Body.Close() }()
-		b, _ := io.ReadAll(resp.Body)
+	recordAPIResponseMetadata(ctx, e.cfg, httpResp.StatusCode, httpResp.Header.Clone())
+	if httpResp.StatusCode < 200 || httpResp.StatusCode >= 300 {
+		b, _ := io.ReadAll(httpResp.Body)
 		appendAPIResponseChunk(ctx, e.cfg, b)
-		log.Debugf("request error, error status: %d, error body: %s", resp.StatusCode, string(b))
-		return nil, statusErr{code: resp.StatusCode, msg: string(b)}
+		log.Debugf("request error, error status: %d, error body: %s", httpResp.StatusCode, string(b))
+		if errClose := httpResp.Body.Close(); errClose != nil {
+			log.Errorf("qwen executor: close response body error: %v", errClose)
+		}
+		err = statusErr{code: httpResp.StatusCode, msg: string(b)}
+		return nil, err
 	}
 	out := make(chan cliproxyexecutor.StreamChunk)
+	stream = out
 	go func() {
 		defer close(out)
-		defer func() { _ = resp.Body.Close() }()
-		scanner := bufio.NewScanner(resp.Body)
+		defer func() {
+			if errClose := httpResp.Body.Close(); errClose != nil {
+				log.Errorf("qwen executor: close response body error: %v", errClose)
+			}
+		}()
+		scanner := bufio.NewScanner(httpResp.Body)
 		buf := make([]byte, 20_971_520)
 		scanner.Buffer(buf, 20_971_520)
 		var param any
@@ -140,11 +193,17 @@ func (e *QwenExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 				out <- cliproxyexecutor.StreamChunk{Payload: []byte(chunks[i])}
 			}
 		}
-		if err = scanner.Err(); err != nil {
-			out <- cliproxyexecutor.StreamChunk{Err: err}
+		doneChunks := sdktranslator.TranslateStream(ctx, to, from, req.Model, bytes.Clone(opts.OriginalRequest), body, bytes.Clone([]byte("[DONE]")), &param)
+		for i := range doneChunks {
+			out <- cliproxyexecutor.StreamChunk{Payload: []byte(doneChunks[i])}
+		}
+		if errScan := scanner.Err(); errScan != nil {
+			recordAPIResponseError(ctx, e.cfg, errScan)
+			reporter.publishFailure(ctx)
+			out <- cliproxyexecutor.StreamChunk{Err: errScan}
 		}
 	}()
-	return out, nil
+	return stream, nil
 }
 
 func (e *QwenExecutor) CountTokens(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (cliproxyexecutor.Response, error) {

internal/runtime/executor/usage_helpers.go

@@ -41,6 +41,23 @@ func newUsageReporter(ctx context.Context, provider, model string, auth *cliprox
 }
 
 func (r *usageReporter) publish(ctx context.Context, detail usage.Detail) {
+	r.publishWithOutcome(ctx, detail, false)
+}
+
+func (r *usageReporter) publishFailure(ctx context.Context) {
+	r.publishWithOutcome(ctx, usage.Detail{}, true)
+}
+
+func (r *usageReporter) trackFailure(ctx context.Context, errPtr *error) {
+	if r == nil || errPtr == nil {
+		return
+	}
+	if *errPtr != nil {
+		r.publishFailure(ctx)
+	}
+}
+
+func (r *usageReporter) publishWithOutcome(ctx context.Context, detail usage.Detail, failed bool) {
 	if r == nil {
 		return
 	}
@@ -50,7 +67,7 @@ func (r *usageReporter) publish(ctx context.Context, detail usage.Detail) {
 			detail.TotalTokens = total
 		}
 	}
-	if detail.InputTokens == 0 && detail.OutputTokens == 0 && detail.ReasoningTokens == 0 && detail.CachedTokens == 0 && detail.TotalTokens == 0 {
+	if detail.InputTokens == 0 && detail.OutputTokens == 0 && detail.ReasoningTokens == 0 && detail.CachedTokens == 0 && detail.TotalTokens == 0 && !failed {
 		return
 	}
 	r.once.Do(func() {
@@ -61,6 +78,7 @@ func (r *usageReporter) publish(ctx context.Context, detail usage.Detail) {
 			APIKey:      r.apiKey,
 			AuthID:      r.authID,
 			RequestedAt: r.requestedAt,
+			Failed:      failed,
 			Detail:      detail,
 		})
 	})

internal/translator/codex/claude/codex_claude_response.go

@@ -7,7 +7,6 @@
 package claude
 
 import (
-	"bufio"
 	"bytes"
 	"context"
 	"encoding/json"
@@ -180,56 +179,58 @@ func ConvertCodexResponseToClaude(_ context.Context, _ string, originalRequestRa
 // Returns:
 //   - string: A Claude Code-compatible JSON response containing all message content and metadata
 func ConvertCodexResponseToClaudeNonStream(_ context.Context, _ string, originalRequestRawJSON, _ []byte, rawJSON []byte, _ *any) string {
-	scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
-	buffer := make([]byte, 20_971_520)
-	scanner.Buffer(buffer, 20_971_520)
 	revNames := buildReverseMapFromClaudeOriginalShortToOriginal(originalRequestRawJSON)
 
-	for scanner.Scan() {
-		line := scanner.Bytes()
-		if !bytes.HasPrefix(line, dataTag) {
-			continue
-		}
-		payload := bytes.TrimSpace(line[len(dataTag):])
-		if len(payload) == 0 {
-			continue
-		}
+	rootResult := gjson.ParseBytes(rawJSON)
+	if rootResult.Get("type").String() != "response.completed" {
+		return ""
+	}
 
-		rootResult := gjson.ParseBytes(payload)
-		if rootResult.Get("type").String() != "response.completed" {
-			continue
-		}
+	responseData := rootResult.Get("response")
+	if !responseData.Exists() {
+		return ""
+	}
 
-		responseData := rootResult.Get("response")
-		if !responseData.Exists() {
-			continue
-		}
+	response := map[string]interface{}{
+		"id":            responseData.Get("id").String(),
+		"type":          "message",
+		"role":          "assistant",
+		"model":         responseData.Get("model").String(),
+		"content":       []interface{}{},
+		"stop_reason":   nil,
+		"stop_sequence": nil,
+		"usage": map[string]interface{}{
+			"input_tokens":  responseData.Get("usage.input_tokens").Int(),
+			"output_tokens": responseData.Get("usage.output_tokens").Int(),
+		},
+	}
 
-		response := map[string]interface{}{
-			"id":            responseData.Get("id").String(),
-			"type":          "message",
-			"role":          "assistant",
-			"model":         responseData.Get("model").String(),
-			"content":       []interface{}{},
-			"stop_reason":   nil,
-			"stop_sequence": nil,
-			"usage": map[string]interface{}{
-				"input_tokens":  responseData.Get("usage.input_tokens").Int(),
-				"output_tokens": responseData.Get("usage.output_tokens").Int(),
-			},
-		}
+	var contentBlocks []interface{}
+	hasToolCall := false
 
-		var contentBlocks []interface{}
-		hasToolCall := false
-
-		if output := responseData.Get("output"); output.Exists() && output.IsArray() {
-			output.ForEach(func(_, item gjson.Result) bool {
-				switch item.Get("type").String() {
-				case "reasoning":
-					thinkingBuilder := strings.Builder{}
-					if summary := item.Get("summary"); summary.Exists() {
-						if summary.IsArray() {
-							summary.ForEach(func(_, part gjson.Result) bool {
+	if output := responseData.Get("output"); output.Exists() && output.IsArray() {
+		output.ForEach(func(_, item gjson.Result) bool {
+			switch item.Get("type").String() {
+			case "reasoning":
+				thinkingBuilder := strings.Builder{}
+				if summary := item.Get("summary"); summary.Exists() {
+					if summary.IsArray() {
+						summary.ForEach(func(_, part gjson.Result) bool {
+							if txt := part.Get("text"); txt.Exists() {
+								thinkingBuilder.WriteString(txt.String())
+							} else {
+								thinkingBuilder.WriteString(part.String())
+							}
+							return true
+						})
+					} else {
+						thinkingBuilder.WriteString(summary.String())
+					}
+				}
+				if thinkingBuilder.Len() == 0 {
+					if content := item.Get("content"); content.Exists() {
+						if content.IsArray() {
+							content.ForEach(func(_, part gjson.Result) bool {
 								if txt := part.Get("text"); txt.Exists() {
 									thinkingBuilder.WriteString(txt.String())
 								} else {
@@ -238,114 +239,96 @@ func ConvertCodexResponseToClaudeNonStream(_ context.Context, _ string, original
 								return true
 							})
 						} else {
-							thinkingBuilder.WriteString(summary.String())
+							thinkingBuilder.WriteString(content.String())
 						}
 					}
-					if thinkingBuilder.Len() == 0 {
-						if content := item.Get("content"); content.Exists() {
-							if content.IsArray() {
-								content.ForEach(func(_, part gjson.Result) bool {
-									if txt := part.Get("text"); txt.Exists() {
-										thinkingBuilder.WriteString(txt.String())
-									} else {
-										thinkingBuilder.WriteString(part.String())
-									}
-									return true
-								})
-							} else {
-								thinkingBuilder.WriteString(content.String())
-							}
-						}
-					}
-					if thinkingBuilder.Len() > 0 {
-						contentBlocks = append(contentBlocks, map[string]interface{}{
-							"type":     "thinking",
-							"thinking": thinkingBuilder.String(),
-						})
-					}
-				case "message":
-					if content := item.Get("content"); content.Exists() {
-						if content.IsArray() {
-							content.ForEach(func(_, part gjson.Result) bool {
-								if part.Get("type").String() == "output_text" {
-									text := part.Get("text").String()
-									if text != "" {
-										contentBlocks = append(contentBlocks, map[string]interface{}{
-											"type": "text",
-											"text": text,
-										})
-									}
-								}
-								return true
-							})
-						} else {
-							text := content.String()
-							if text != "" {
-								contentBlocks = append(contentBlocks, map[string]interface{}{
-									"type": "text",
-									"text": text,
-								})
-							}
-						}
-					}
-				case "function_call":
-					hasToolCall = true
-					name := item.Get("name").String()
-					if original, ok := revNames[name]; ok {
-						name = original
-					}
-
-					toolBlock := map[string]interface{}{
-						"type":  "tool_use",
-						"id":    item.Get("call_id").String(),
-						"name":  name,
-						"input": map[string]interface{}{},
-					}
-
-					if argsStr := item.Get("arguments").String(); argsStr != "" {
-						var args interface{}
-						if err := json.Unmarshal([]byte(argsStr), &args); err == nil {
-							toolBlock["input"] = args
-						}
-					}
-
-					contentBlocks = append(contentBlocks, toolBlock)
 				}
-				return true
-			})
-		}
+				if thinkingBuilder.Len() > 0 {
+					contentBlocks = append(contentBlocks, map[string]interface{}{
+						"type":     "thinking",
+						"thinking": thinkingBuilder.String(),
+					})
+				}
+			case "message":
+				if content := item.Get("content"); content.Exists() {
+					if content.IsArray() {
+						content.ForEach(func(_, part gjson.Result) bool {
+							if part.Get("type").String() == "output_text" {
+								text := part.Get("text").String()
+								if text != "" {
+									contentBlocks = append(contentBlocks, map[string]interface{}{
+										"type": "text",
+										"text": text,
+									})
+								}
+							}
+							return true
+						})
+					} else {
+						text := content.String()
+						if text != "" {
+							contentBlocks = append(contentBlocks, map[string]interface{}{
+								"type": "text",
+								"text": text,
+							})
+						}
+					}
+				}
+			case "function_call":
+				hasToolCall = true
+				name := item.Get("name").String()
+				if original, ok := revNames[name]; ok {
+					name = original
+				}
 
-		if len(contentBlocks) > 0 {
-			response["content"] = contentBlocks
-		}
+				toolBlock := map[string]interface{}{
+					"type":  "tool_use",
+					"id":    item.Get("call_id").String(),
+					"name":  name,
+					"input": map[string]interface{}{},
+				}
 
-		if stopReason := responseData.Get("stop_reason"); stopReason.Exists() && stopReason.String() != "" {
-			response["stop_reason"] = stopReason.String()
-		} else if hasToolCall {
-			response["stop_reason"] = "tool_use"
-		} else {
-			response["stop_reason"] = "end_turn"
-		}
+				if argsStr := item.Get("arguments").String(); argsStr != "" {
+					var args interface{}
+					if err := json.Unmarshal([]byte(argsStr), &args); err == nil {
+						toolBlock["input"] = args
+					}
+				}
 
-		if stopSequence := responseData.Get("stop_sequence"); stopSequence.Exists() && stopSequence.String() != "" {
-			response["stop_sequence"] = stopSequence.Value()
-		}
-
-		if responseData.Get("usage.input_tokens").Exists() || responseData.Get("usage.output_tokens").Exists() {
-			response["usage"] = map[string]interface{}{
-				"input_tokens":  responseData.Get("usage.input_tokens").Int(),
-				"output_tokens": responseData.Get("usage.output_tokens").Int(),
+				contentBlocks = append(contentBlocks, toolBlock)
 			}
-		}
-
-		responseJSON, err := json.Marshal(response)
-		if err != nil {
-			return ""
-		}
-		return string(responseJSON)
+			return true
+		})
 	}
 
-	return ""
+	if len(contentBlocks) > 0 {
+		response["content"] = contentBlocks
+	}
+
+	if stopReason := responseData.Get("stop_reason"); stopReason.Exists() && stopReason.String() != "" {
+		response["stop_reason"] = stopReason.String()
+	} else if hasToolCall {
+		response["stop_reason"] = "tool_use"
+	} else {
+		response["stop_reason"] = "end_turn"
+	}
+
+	if stopSequence := responseData.Get("stop_sequence"); stopSequence.Exists() && stopSequence.String() != "" {
+		response["stop_sequence"] = stopSequence.Value()
+	}
+
+	if responseData.Get("usage.input_tokens").Exists() || responseData.Get("usage.output_tokens").Exists() {
+		response["usage"] = map[string]interface{}{
+			"input_tokens":  responseData.Get("usage.input_tokens").Int(),
+			"output_tokens": responseData.Get("usage.output_tokens").Int(),
+		}
+	}
+
+	responseJSON, err := json.Marshal(response)
+	if err != nil {
+		return ""
+	}
+	return string(responseJSON)
 }
 
 // buildReverseMapFromClaudeOriginalShortToOriginal builds a map[short]original from original Claude request tools.

internal/translator/codex/gemini/codex_gemini_response.go

@@ -5,7 +5,6 @@
 package gemini
 
 import (
-	"bufio"
 	"bytes"
 	"context"
 	"encoding/json"
@@ -152,159 +151,146 @@ func ConvertCodexResponseToGemini(_ context.Context, modelName string, originalR
 // Returns:
 //   - string: A Gemini-compatible JSON response containing all message content and metadata
 func ConvertCodexResponseToGeminiNonStream(_ context.Context, modelName string, originalRequestRawJSON, requestRawJSON, rawJSON []byte, _ *any) string {
-	scanner := bufio.NewScanner(bytes.NewReader(rawJSON))
-	buffer := make([]byte, 20_971_520)
-	scanner.Buffer(buffer, 20_971_520)
-	for scanner.Scan() {
-		line := scanner.Bytes()
-		// log.Debug(string(line))
-		if !bytes.HasPrefix(line, dataTag) {
-			continue
-		}
-		rawJSON = bytes.TrimSpace(rawJSON[5:])
+	rootResult := gjson.ParseBytes(rawJSON)
 
-		rootResult := gjson.ParseBytes(rawJSON)
-
-		// Verify this is a response.completed event
-		if rootResult.Get("type").String() != "response.completed" {
-			continue
-		}
-
-		// Base Gemini response template for non-streaming
-		template := `{"candidates":[{"content":{"role":"model","parts":[]},"finishReason":"STOP"}],"usageMetadata":{"trafficType":"PROVISIONED_THROUGHPUT"},"modelVersion":"","createTime":"","responseId":""}`
-
-		// Set model version
-		template, _ = sjson.Set(template, "modelVersion", modelName)
-
-		// Set response metadata from the completed response
-		responseData := rootResult.Get("response")
-		if responseData.Exists() {
-			// Set response ID
-			if responseId := responseData.Get("id"); responseId.Exists() {
-				template, _ = sjson.Set(template, "responseId", responseId.String())
-			}
-
-			// Set creation time
-			if createdAt := responseData.Get("created_at"); createdAt.Exists() {
-				template, _ = sjson.Set(template, "createTime", time.Unix(createdAt.Int(), 0).Format(time.RFC3339Nano))
-			}
-
-			// Set usage metadata
-			if usage := responseData.Get("usage"); usage.Exists() {
-				inputTokens := usage.Get("input_tokens").Int()
-				outputTokens := usage.Get("output_tokens").Int()
-				totalTokens := inputTokens + outputTokens
-
-				template, _ = sjson.Set(template, "usageMetadata.promptTokenCount", inputTokens)
-				template, _ = sjson.Set(template, "usageMetadata.candidatesTokenCount", outputTokens)
-				template, _ = sjson.Set(template, "usageMetadata.totalTokenCount", totalTokens)
-			}
-
-			// Process output content to build parts array
-			var parts []interface{}
-			hasToolCall := false
-			var pendingFunctionCalls []interface{}
-
-			flushPendingFunctionCalls := func() {
-				if len(pendingFunctionCalls) > 0 {
-					// Add all pending function calls as individual parts
-					// This maintains the original Gemini API format while ensuring consecutive calls are grouped together
-					for _, fc := range pendingFunctionCalls {
-						parts = append(parts, fc)
-					}
-					pendingFunctionCalls = nil
-				}
-			}
-
-			if output := responseData.Get("output"); output.Exists() && output.IsArray() {
-				output.ForEach(func(key, value gjson.Result) bool {
-					itemType := value.Get("type").String()
-
-					switch itemType {
-					case "reasoning":
-						// Flush any pending function calls before adding non-function content
-						flushPendingFunctionCalls()
-
-						// Add thinking content
-						if content := value.Get("content"); content.Exists() {
-							part := map[string]interface{}{
-								"thought": true,
-								"text":    content.String(),
-							}
-							parts = append(parts, part)
-						}
-
-					case "message":
-						// Flush any pending function calls before adding non-function content
-						flushPendingFunctionCalls()
-
-						// Add regular text content
-						if content := value.Get("content"); content.Exists() && content.IsArray() {
-							content.ForEach(func(_, contentItem gjson.Result) bool {
-								if contentItem.Get("type").String() == "output_text" {
-									if text := contentItem.Get("text"); text.Exists() {
-										part := map[string]interface{}{
-											"text": text.String(),
-										}
-										parts = append(parts, part)
-									}
-								}
-								return true
-							})
-						}
-
-					case "function_call":
-						// Collect function call for potential merging with consecutive ones
-						hasToolCall = true
-						functionCall := map[string]interface{}{
-							"functionCall": map[string]interface{}{
-								"name": func() string {
-									n := value.Get("name").String()
-									rev := buildReverseMapFromGeminiOriginal(originalRequestRawJSON)
-									if orig, ok := rev[n]; ok {
-										return orig
-									}
-									return n
-								}(),
-								"args": map[string]interface{}{},
-							},
-						}
-
-						// Parse and set arguments
-						if argsStr := value.Get("arguments").String(); argsStr != "" {
-							argsResult := gjson.Parse(argsStr)
-							if argsResult.IsObject() {
-								var args map[string]interface{}
-								if err := json.Unmarshal([]byte(argsStr), &args); err == nil {
-									functionCall["functionCall"].(map[string]interface{})["args"] = args
-								}
-							}
-						}
-
-						pendingFunctionCalls = append(pendingFunctionCalls, functionCall)
-					}
-					return true
-				})
-
-				// Handle any remaining pending function calls at the end
-				flushPendingFunctionCalls()
-			}
-
-			// Set the parts array
-			if len(parts) > 0 {
-				template, _ = sjson.SetRaw(template, "candidates.0.content.parts", mustMarshalJSON(parts))
-			}
-
-			// Set finish reason based on whether there were tool calls
-			if hasToolCall {
-				template, _ = sjson.Set(template, "candidates.0.finishReason", "STOP")
-			} else {
-				template, _ = sjson.Set(template, "candidates.0.finishReason", "STOP")
-			}
-		}
-		return template
+	// Verify this is a response.completed event
+	if rootResult.Get("type").String() != "response.completed" {
+		return ""
 	}
-	return ""
+
+	// Base Gemini response template for non-streaming
+	template := `{"candidates":[{"content":{"role":"model","parts":[]},"finishReason":"STOP"}],"usageMetadata":{"trafficType":"PROVISIONED_THROUGHPUT"},"modelVersion":"","createTime":"","responseId":""}`
+
+	// Set model version
+	template, _ = sjson.Set(template, "modelVersion", modelName)
+
+	// Set response metadata from the completed response
+	responseData := rootResult.Get("response")
+	if responseData.Exists() {
+		// Set response ID
+		if responseId := responseData.Get("id"); responseId.Exists() {
+			template, _ = sjson.Set(template, "responseId", responseId.String())
+		}
+
+		// Set creation time
+		if createdAt := responseData.Get("created_at"); createdAt.Exists() {
+			template, _ = sjson.Set(template, "createTime", time.Unix(createdAt.Int(), 0).Format(time.RFC3339Nano))
+		}
+
+		// Set usage metadata
+		if usage := responseData.Get("usage"); usage.Exists() {
+			inputTokens := usage.Get("input_tokens").Int()
+			outputTokens := usage.Get("output_tokens").Int()
+			totalTokens := inputTokens + outputTokens
+
+			template, _ = sjson.Set(template, "usageMetadata.promptTokenCount", inputTokens)
+			template, _ = sjson.Set(template, "usageMetadata.candidatesTokenCount", outputTokens)
+			template, _ = sjson.Set(template, "usageMetadata.totalTokenCount", totalTokens)
+		}
+
+		// Process output content to build parts array
+		var parts []interface{}
+		hasToolCall := false
+		var pendingFunctionCalls []interface{}
+
+		flushPendingFunctionCalls := func() {
+			if len(pendingFunctionCalls) > 0 {
+				// Add all pending function calls as individual parts
+				// This maintains the original Gemini API format while ensuring consecutive calls are grouped together
+				for _, fc := range pendingFunctionCalls {
+					parts = append(parts, fc)
+				}
+				pendingFunctionCalls = nil
+			}
+		}
+
+		if output := responseData.Get("output"); output.Exists() && output.IsArray() {
+			output.ForEach(func(key, value gjson.Result) bool {
+				itemType := value.Get("type").String()
+
+				switch itemType {
+				case "reasoning":
+					// Flush any pending function calls before adding non-function content
+					flushPendingFunctionCalls()
+
+					// Add thinking content
+					if content := value.Get("content"); content.Exists() {
+						part := map[string]interface{}{
+							"thought": true,
+							"text":    content.String(),
+						}
+						parts = append(parts, part)
+					}
+
+				case "message":
+					// Flush any pending function calls before adding non-function content
+					flushPendingFunctionCalls()
+
+					// Add regular text content
+					if content := value.Get("content"); content.Exists() && content.IsArray() {
+						content.ForEach(func(_, contentItem gjson.Result) bool {
+							if contentItem.Get("type").String() == "output_text" {
+								if text := contentItem.Get("text"); text.Exists() {
+									part := map[string]interface{}{
+										"text": text.String(),
+									}
+									parts = append(parts, part)
+								}
+							}
+							return true
+						})
+					}
+
+				case "function_call":
+					// Collect function call for potential merging with consecutive ones
+					hasToolCall = true
+					functionCall := map[string]interface{}{
+						"functionCall": map[string]interface{}{
+							"name": func() string {
+								n := value.Get("name").String()
+								rev := buildReverseMapFromGeminiOriginal(originalRequestRawJSON)
+								if orig, ok := rev[n]; ok {
+									return orig
+								}
+								return n
+							}(),
+							"args": map[string]interface{}{},
+						},
+					}
+
+					// Parse and set arguments
+					if argsStr := value.Get("arguments").String(); argsStr != "" {
+						argsResult := gjson.Parse(argsStr)
+						if argsResult.IsObject() {
+							var args map[string]interface{}
+							if err := json.Unmarshal([]byte(argsStr), &args); err == nil {
+								functionCall["functionCall"].(map[string]interface{})["args"] = args
+							}
+						}
+					}
+
+					pendingFunctionCalls = append(pendingFunctionCalls, functionCall)
+				}
+				return true
+			})
+
+			// Handle any remaining pending function calls at the end
+			flushPendingFunctionCalls()
+		}
+
+		// Set the parts array
+		if len(parts) > 0 {
+			template, _ = sjson.SetRaw(template, "candidates.0.content.parts", mustMarshalJSON(parts))
+		}
+
+		// Set finish reason based on whether there were tool calls
+		if hasToolCall {
+			template, _ = sjson.Set(template, "candidates.0.finishReason", "STOP")
+		} else {
+			template, _ = sjson.Set(template, "candidates.0.finishReason", "STOP")
+		}
+	}
+	return template
 }
 
 // buildReverseMapFromGeminiOriginal builds a map[short]original from original Gemini request tools.

internal/translator/gemini-cli/claude/gemini-cli_claude_request.go

@@ -11,7 +11,6 @@ import (
 	"strings"
 
 	client "github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
-	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -36,18 +35,6 @@ import (
 //   - []byte: The transformed request data in Gemini CLI API format
 func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []byte {
 	rawJSON := bytes.Clone(inputRawJSON)
-	var pathsToDelete []string
-	root := gjson.ParseBytes(rawJSON)
-	util.Walk(root, "", "additionalProperties", &pathsToDelete)
-	util.Walk(root, "", "$schema", &pathsToDelete)
-
-	var err error
-	for _, p := range pathsToDelete {
-		rawJSON, err = sjson.DeleteBytes(rawJSON, p)
-		if err != nil {
-			continue
-		}
-	}
 	rawJSON = bytes.Replace(rawJSON, []byte(`"url":{"type":"string","format":"uri",`), []byte(`"url":{"type":"string",`), -1)
 
 	// system instruction
@@ -99,7 +86,7 @@ func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []
 						functionName := contentResult.Get("name").String()
 						functionArgs := contentResult.Get("input").String()
 						var args map[string]any
-						if err = json.Unmarshal([]byte(functionArgs), &args); err == nil {
+						if err := json.Unmarshal([]byte(functionArgs), &args); err == nil {
 							clientContent.Parts = append(clientContent.Parts, client.Part{FunctionCall: &client.FunctionCall{Name: functionName, Args: args}})
 						}
 					} else if contentTypeResult.Type == gjson.String && contentTypeResult.String() == "tool_result" {
@@ -136,18 +123,10 @@ func ConvertClaudeRequestToCLI(modelName string, inputRawJSON []byte, _ bool) []
 			inputSchemaResult := toolResult.Get("input_schema")
 			if inputSchemaResult.Exists() && inputSchemaResult.IsObject() {
 				inputSchema := inputSchemaResult.Raw
-				// Use comprehensive schema sanitization for Gemini API compatibility
-				if sanitizedSchema, sanitizeErr := util.SanitizeSchemaForGemini(inputSchema); sanitizeErr == nil {
-					inputSchema = sanitizedSchema
-				} else {
-					// Fallback to basic cleanup if sanitization fails
-					inputSchema, _ = sjson.Delete(inputSchema, "additionalProperties")
-					inputSchema, _ = sjson.Delete(inputSchema, "$schema")
-				}
 				tool, _ := sjson.Delete(toolResult.Raw, "input_schema")
-				tool, _ = sjson.SetRaw(tool, "parameters", inputSchema)
+				tool, _ = sjson.SetRaw(tool, "parametersJsonSchema", inputSchema)
 				var toolDeclaration any
-				if err = json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
+				if err := json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
 					tools[0].FunctionDeclarations = append(tools[0].FunctionDeclarations, toolDeclaration)
 				}
 			}

internal/translator/gemini-cli/gemini/gemini-cli_gemini_request.go

@@ -10,6 +10,7 @@ import (
 	"encoding/json"
 	"fmt"
 
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	log "github.com/sirupsen/logrus"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
@@ -78,6 +79,24 @@ func ConvertGeminiRequestToGeminiCLI(_ string, inputRawJSON []byte, _ bool) []by
 		})
 	}
 
+	toolsResult := gjson.GetBytes(rawJSON, "request.tools")
+	if toolsResult.Exists() && toolsResult.IsArray() {
+		toolResults := toolsResult.Array()
+		for i := 0; i < len(toolResults); i++ {
+			functionDeclarationsResult := gjson.GetBytes(rawJSON, fmt.Sprintf("request.tools.%d.function_declarations", i))
+			if functionDeclarationsResult.Exists() && functionDeclarationsResult.IsArray() {
+				functionDeclarationsResults := functionDeclarationsResult.Array()
+				for j := 0; j < len(functionDeclarationsResults); j++ {
+					parametersResult := gjson.GetBytes(rawJSON, fmt.Sprintf("request.tools.%d.function_declarations.%d.parameters", i, j))
+					if parametersResult.Exists() {
+						strJson, _ := util.RenameKey(string(rawJSON), fmt.Sprintf("request.tools.%d.function_declarations.%d.parameters", i, j), fmt.Sprintf("request.tools.%d.function_declarations.%d.parametersJsonSchema", i, j))
+						rawJSON = []byte(strJson)
+					}
+				}
+			}
+		}
+	}
+
 	return rawJSON
 }
 

internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_request.go

@@ -26,21 +26,6 @@ import (
 //   - []byte: The transformed request data in Gemini CLI API format
 func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bool) []byte {
 	rawJSON := bytes.Clone(inputRawJSON)
-	var pathsToDelete []string
-	root := gjson.ParseBytes(rawJSON)
-	util.Walk(root, "", "additionalProperties", &pathsToDelete)
-	util.Walk(root, "", "$schema", &pathsToDelete)
-	util.Walk(root, "", "ref", &pathsToDelete)
-	util.Walk(root, "", "strict", &pathsToDelete)
-
-	var err error
-	for _, p := range pathsToDelete {
-		rawJSON, err = sjson.DeleteBytes(rawJSON, p)
-		if err != nil {
-			continue
-		}
-	}
-
 	// Base envelope
 	out := []byte(`{"project":"","request":{"contents":[],"generationConfig":{"thinkingConfig":{"include_thoughts":true}}},"model":"gemini-2.5-pro"}`)
 
@@ -265,22 +250,13 @@ func ConvertOpenAIRequestToGeminiCLI(modelName string, inputRawJSON []byte, _ bo
 			if t.Get("type").String() == "function" {
 				fn := t.Get("function")
 				if fn.Exists() && fn.IsObject() {
-					out, _ = sjson.SetRawBytes(out, fdPath+".-1", []byte(fn.Raw))
+					parametersJsonSchema, _ := util.RenameKey(fn.Raw, "parameters", "parametersJsonSchema")
+					out, _ = sjson.SetRawBytes(out, fdPath+".-1", []byte(parametersJsonSchema))
 				}
 			}
 		}
 	}
 
-	var pathsToType []string
-	root = gjson.ParseBytes(out)
-	util.Walk(root, "", "type", &pathsToType)
-	for _, p := range pathsToType {
-		typeResult := gjson.GetBytes(out, p)
-		if strings.ToLower(typeResult.String()) == "select" {
-			out, _ = sjson.SetBytes(out, p, "STRING")
-		}
-	}
-
 	return out
 }
 

internal/translator/gemini-cli/openai/chat-completions/gemini-cli_openai_response.go

@@ -97,6 +97,7 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
 
 	// Process the main content part of the response.
 	partsResult := gjson.GetBytes(rawJSON, "response.candidates.0.content.parts")
+	hasFunctionCall := false
 	if partsResult.IsArray() {
 		partResults := partsResult.Array()
 		for i := 0; i < len(partResults); i++ {
@@ -118,6 +119,7 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
 				template, _ = sjson.Set(template, "choices.0.delta.role", "assistant")
 			} else if functionCallResult.Exists() {
 				// Handle function call content.
+				hasFunctionCall = true
 				toolCallsResult := gjson.Get(template, "choices.0.delta.tool_calls")
 				functionCallIndex := (*param).(*convertCliResponseToOpenAIChatParams).FunctionIndex
 				(*param).(*convertCliResponseToOpenAIChatParams).FunctionIndex++
@@ -169,6 +171,11 @@ func ConvertCliResponseToOpenAI(_ context.Context, _ string, originalRequestRawJ
 		}
 	}
 
+	if hasFunctionCall {
+		template, _ = sjson.Set(template, "choices.0.finish_reason", "tool_calls")
+		template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
+	}
+
 	return []string{template}
 }
 

internal/translator/gemini/claude/gemini_claude_request.go

@@ -11,7 +11,6 @@ import (
 	"strings"
 
 	client "github.com/router-for-me/CLIProxyAPI/v6/internal/interfaces"
-	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -29,18 +28,6 @@ import (
 //   - []byte: The transformed request in Gemini CLI format.
 func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool) []byte {
 	rawJSON := bytes.Clone(inputRawJSON)
-	var pathsToDelete []string
-	root := gjson.ParseBytes(rawJSON)
-	util.Walk(root, "", "additionalProperties", &pathsToDelete)
-	util.Walk(root, "", "$schema", &pathsToDelete)
-
-	var err error
-	for _, p := range pathsToDelete {
-		rawJSON, err = sjson.DeleteBytes(rawJSON, p)
-		if err != nil {
-			continue
-		}
-	}
 	rawJSON = bytes.Replace(rawJSON, []byte(`"url":{"type":"string","format":"uri",`), []byte(`"url":{"type":"string",`), -1)
 
 	// system instruction
@@ -92,7 +79,7 @@ func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 						functionName := contentResult.Get("name").String()
 						functionArgs := contentResult.Get("input").String()
 						var args map[string]any
-						if err = json.Unmarshal([]byte(functionArgs), &args); err == nil {
+						if err := json.Unmarshal([]byte(functionArgs), &args); err == nil {
 							clientContent.Parts = append(clientContent.Parts, client.Part{FunctionCall: &client.FunctionCall{Name: functionName, Args: args}})
 						}
 					} else if contentTypeResult.Type == gjson.String && contentTypeResult.String() == "tool_result" {
@@ -129,18 +116,10 @@ func ConvertClaudeRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 			inputSchemaResult := toolResult.Get("input_schema")
 			if inputSchemaResult.Exists() && inputSchemaResult.IsObject() {
 				inputSchema := inputSchemaResult.Raw
-				// Use comprehensive schema sanitization for Gemini API compatibility
-				if sanitizedSchema, sanitizeErr := util.SanitizeSchemaForGemini(inputSchema); sanitizeErr == nil {
-					inputSchema = sanitizedSchema
-				} else {
-					// Fallback to basic cleanup if sanitization fails
-					inputSchema, _ = sjson.Delete(inputSchema, "additionalProperties")
-					inputSchema, _ = sjson.Delete(inputSchema, "$schema")
-				}
 				tool, _ := sjson.Delete(toolResult.Raw, "input_schema")
-				tool, _ = sjson.SetRaw(tool, "parameters", inputSchema)
+				tool, _ = sjson.SetRaw(tool, "parametersJsonSchema", inputSchema)
 				var toolDeclaration any
-				if err = json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
+				if err := json.Unmarshal([]byte(tool), &toolDeclaration); err == nil {
 					tools[0].FunctionDeclarations = append(tools[0].FunctionDeclarations, toolDeclaration)
 				}
 			}

internal/translator/gemini/gemini-cli/gemini_gemini-cli_request.go

@@ -7,7 +7,9 @@ package geminiCLI
 
 import (
 	"bytes"
+	"fmt"
 
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -24,5 +26,24 @@ func ConvertGeminiCLIRequestToGemini(_ string, inputRawJSON []byte, _ bool) []by
 		rawJSON, _ = sjson.SetRawBytes(rawJSON, "system_instruction", []byte(gjson.GetBytes(rawJSON, "systemInstruction").Raw))
 		rawJSON, _ = sjson.DeleteBytes(rawJSON, "systemInstruction")
 	}
+
+	toolsResult := gjson.GetBytes(rawJSON, "tools")
+	if toolsResult.Exists() && toolsResult.IsArray() {
+		toolResults := toolsResult.Array()
+		for i := 0; i < len(toolResults); i++ {
+			functionDeclarationsResult := gjson.GetBytes(rawJSON, fmt.Sprintf("tools.%d.function_declarations", i))
+			if functionDeclarationsResult.Exists() && functionDeclarationsResult.IsArray() {
+				functionDeclarationsResults := functionDeclarationsResult.Array()
+				for j := 0; j < len(functionDeclarationsResults); j++ {
+					parametersResult := gjson.GetBytes(rawJSON, fmt.Sprintf("tools.%d.function_declarations.%d.parameters", i, j))
+					if parametersResult.Exists() {
+						strJson, _ := util.RenameKey(string(rawJSON), fmt.Sprintf("tools.%d.function_declarations.%d.parameters", i, j), fmt.Sprintf("tools.%d.function_declarations.%d.parametersJsonSchema", i, j))
+						rawJSON = []byte(strJson)
+					}
+				}
+			}
+		}
+	}
+
 	return rawJSON
 }

internal/translator/gemini/gemini/gemini_gemini_request.go

@@ -7,6 +7,7 @@ import (
 	"bytes"
 	"fmt"
 
+	"github.com/router-for-me/CLIProxyAPI/v6/internal/util"
 	"github.com/tidwall/gjson"
 	"github.com/tidwall/sjson"
 )
@@ -24,6 +25,24 @@ func ConvertGeminiRequestToGemini(_ string, inputRawJSON []byte, _ bool) []byte
 		return rawJSON
 	}
 
+	toolsResult := gjson.GetBytes(rawJSON, "tools")
+	if toolsResult.Exists() && toolsResult.IsArray() {
+		toolResults := toolsResult.Array()
+		for i := 0; i < len(toolResults); i++ {
+			functionDeclarationsResult := gjson.GetBytes(rawJSON, fmt.Sprintf("tools.%d.function_declarations", i))
+			if functionDeclarationsResult.Exists() && functionDeclarationsResult.IsArray() {
+				functionDeclarationsResults := functionDeclarationsResult.Array()
+				for j := 0; j < len(functionDeclarationsResults); j++ {
+					parametersResult := gjson.GetBytes(rawJSON, fmt.Sprintf("tools.%d.function_declarations.%d.parameters", i, j))
+					if parametersResult.Exists() {
+						strJson, _ := util.RenameKey(string(rawJSON), fmt.Sprintf("tools.%d.function_declarations.%d.parameters", i, j), fmt.Sprintf("tools.%d.function_declarations.%d.parametersJsonSchema", i, j))
+						rawJSON = []byte(strJson)
+					}
+				}
+			}
+		}
+	}
+
 	// Walk contents and fix roles
 	out := rawJSON
 	prevRole := ""

internal/translator/gemini/openai/chat-completions/gemini_openai_request.go

@@ -26,21 +26,6 @@ import (
 //   - []byte: The transformed request data in Gemini API format
 func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool) []byte {
 	rawJSON := bytes.Clone(inputRawJSON)
-	var pathsToDelete []string
-	root := gjson.ParseBytes(rawJSON)
-	util.Walk(root, "", "additionalProperties", &pathsToDelete)
-	util.Walk(root, "", "$schema", &pathsToDelete)
-	util.Walk(root, "", "ref", &pathsToDelete)
-	util.Walk(root, "", "strict", &pathsToDelete)
-
-	var err error
-	for _, p := range pathsToDelete {
-		rawJSON, err = sjson.DeleteBytes(rawJSON, p)
-		if err != nil {
-			continue
-		}
-	}
-
 	// Base envelope
 	out := []byte(`{"contents":[],"generationConfig":{"thinkingConfig":{"include_thoughts":true}}}`)
 
@@ -290,22 +275,13 @@ func ConvertOpenAIRequestToGemini(modelName string, inputRawJSON []byte, _ bool)
 			if t.Get("type").String() == "function" {
 				fn := t.Get("function")
 				if fn.Exists() && fn.IsObject() {
-					out, _ = sjson.SetRawBytes(out, fdPath+".-1", []byte(fn.Raw))
+					parametersJsonSchema, _ := util.RenameKey(fn.Raw, "parameters", "parametersJsonSchema")
+					out, _ = sjson.SetRawBytes(out, fdPath+".-1", []byte(parametersJsonSchema))
 				}
 			}
 		}
 	}
 
-	var pathsToType []string
-	root = gjson.ParseBytes(out)
-	util.Walk(root, "", "type", &pathsToType)
-	for _, p := range pathsToType {
-		typeResult := gjson.GetBytes(out, p)
-		if strings.ToLower(typeResult.String()) == "select" {
-			out, _ = sjson.SetBytes(out, p, "STRING")
-		}
-	}
-
 	return out
 }
 

internal/translator/gemini/openai/chat-completions/gemini_openai_response.go

@@ -100,6 +100,7 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
 
 	// Process the main content part of the response.
 	partsResult := gjson.GetBytes(rawJSON, "candidates.0.content.parts")
+	hasFunctionCall := false
 	if partsResult.IsArray() {
 		partResults := partsResult.Array()
 		for i := 0; i < len(partResults); i++ {
@@ -121,6 +122,7 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
 				template, _ = sjson.Set(template, "choices.0.delta.role", "assistant")
 			} else if functionCallResult.Exists() {
 				// Handle function call content.
+				hasFunctionCall = true
 				toolCallsResult := gjson.Get(template, "choices.0.delta.tool_calls")
 				functionCallIndex := (*param).(*convertGeminiResponseToOpenAIChatParams).FunctionIndex
 				(*param).(*convertGeminiResponseToOpenAIChatParams).FunctionIndex++
@@ -172,6 +174,11 @@ func ConvertGeminiResponseToOpenAI(_ context.Context, _ string, originalRequestR
 		}
 	}
 
+	if hasFunctionCall {
+		template, _ = sjson.Set(template, "choices.0.finish_reason", "tool_calls")
+		template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
+	}
+
 	return []string{template}
 }
 
@@ -231,6 +238,7 @@ func ConvertGeminiResponseToOpenAINonStream(_ context.Context, _ string, origina
 
 	// Process the main content part of the response.
 	partsResult := gjson.GetBytes(rawJSON, "candidates.0.content.parts")
+	hasFunctionCall := false
 	if partsResult.IsArray() {
 		partsResults := partsResult.Array()
 		for i := 0; i < len(partsResults); i++ {
@@ -252,6 +260,7 @@ func ConvertGeminiResponseToOpenAINonStream(_ context.Context, _ string, origina
 				template, _ = sjson.Set(template, "choices.0.message.role", "assistant")
 			} else if functionCallResult.Exists() {
 				// Append function call content to the tool_calls array.
+				hasFunctionCall = true
 				toolCallsResult := gjson.Get(template, "choices.0.message.tool_calls")
 				if !toolCallsResult.Exists() || !toolCallsResult.IsArray() {
 					template, _ = sjson.SetRaw(template, "choices.0.message.tool_calls", `[]`)
@@ -297,5 +306,10 @@ func ConvertGeminiResponseToOpenAINonStream(_ context.Context, _ string, origina
 		}
 	}
 
+	if hasFunctionCall {
+		template, _ = sjson.Set(template, "choices.0.finish_reason", "tool_calls")
+		template, _ = sjson.Set(template, "choices.0.native_finish_reason", "tool_calls")
+	}
+
 	return template
 }

internal/translator/gemini/openai/responses/gemini_openai-responses_request.go

@@ -150,7 +150,7 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 					if outputResult.IsObject() {
 						functionResponse, _ = sjson.SetRaw(functionResponse, "functionResponse.response.content", outputResult.String())
 					} else {
-						functionResponse, _ = sjson.Set(functionResponse, "functionResponse.response.content", outputResult.String())
+						functionResponse, _ = sjson.Set(functionResponse, "functionResponse.response.content", output)
 					}
 				}
 
@@ -168,7 +168,7 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 
 		tools.ForEach(func(_, tool gjson.Result) bool {
 			if tool.Get("type").String() == "function" {
-				funcDecl := `{"name":"","description":"","parameters":{}}`
+				funcDecl := `{"name":"","description":"","parametersJsonSchema":{}}`
 
 				if name := tool.Get("name"); name.Exists() {
 					funcDecl, _ = sjson.Set(funcDecl, "name", name.String())
@@ -192,7 +192,7 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 					}
 					// Set the overall type to OBJECT
 					cleaned, _ = sjson.Set(cleaned, "type", "OBJECT")
-					funcDecl, _ = sjson.SetRaw(funcDecl, "parameters", cleaned)
+					funcDecl, _ = sjson.SetRaw(funcDecl, "parametersJsonSchema", cleaned)
 				}
 
 				geminiTools, _ = sjson.SetRaw(geminiTools, "0.functionDeclarations.-1", funcDecl)
@@ -261,6 +261,5 @@ func ConvertOpenAIResponsesRequestToGemini(modelName string, inputRawJSON []byte
 			out, _ = sjson.Set(out, "generationConfig.thinkingConfig.thinkingBudget", -1)
 		}
 	}
-
 	return []byte(out)
 }

internal/translator/openai/claude/openai_claude_response.go

@@ -466,7 +466,7 @@ func ConvertOpenAIResponseToClaudeNonStream(_ context.Context, _ string, origina
 		},
 	}
 
-	var contentBlocks []interface{}
+	contentBlocks := make([]interface{}, 0)
 	hasToolCall := false
 
 	if choices := root.Get("choices"); choices.Exists() && choices.IsArray() && len(choices.Array()) > 0 {
@@ -477,80 +477,90 @@ func ConvertOpenAIResponseToClaudeNonStream(_ context.Context, _ string, origina
 		}
 
 		if message := choice.Get("message"); message.Exists() {
-			if contentArray := message.Get("content"); contentArray.Exists() && contentArray.IsArray() {
-				var textBuilder strings.Builder
-				var thinkingBuilder strings.Builder
+			if contentResult := message.Get("content"); contentResult.Exists() {
+				if contentResult.IsArray() {
+					var textBuilder strings.Builder
+					var thinkingBuilder strings.Builder
 
-				flushText := func() {
-					if textBuilder.Len() == 0 {
-						return
+					flushText := func() {
+						if textBuilder.Len() == 0 {
+							return
+						}
+						contentBlocks = append(contentBlocks, map[string]interface{}{
+							"type": "text",
+							"text": textBuilder.String(),
+						})
+						textBuilder.Reset()
 					}
-					contentBlocks = append(contentBlocks, map[string]interface{}{
-						"type": "text",
-						"text": textBuilder.String(),
-					})
-					textBuilder.Reset()
-				}
 
-				flushThinking := func() {
-					if thinkingBuilder.Len() == 0 {
-						return
+					flushThinking := func() {
+						if thinkingBuilder.Len() == 0 {
+							return
+						}
+						contentBlocks = append(contentBlocks, map[string]interface{}{
+							"type":     "thinking",
+							"thinking": thinkingBuilder.String(),
+						})
+						thinkingBuilder.Reset()
 					}
-					contentBlocks = append(contentBlocks, map[string]interface{}{
-						"type":     "thinking",
-						"thinking": thinkingBuilder.String(),
-					})
-					thinkingBuilder.Reset()
-				}
 
-				for _, item := range contentArray.Array() {
-					typeStr := item.Get("type").String()
-					switch typeStr {
-					case "text":
-						flushThinking()
-						textBuilder.WriteString(item.Get("text").String())
-					case "tool_calls":
-						flushThinking()
-						flushText()
-						toolCalls := item.Get("tool_calls")
-						if toolCalls.IsArray() {
-							toolCalls.ForEach(func(_, tc gjson.Result) bool {
-								hasToolCall = true
-								toolUse := map[string]interface{}{
-									"type": "tool_use",
-									"id":   tc.Get("id").String(),
-									"name": tc.Get("function.name").String(),
-								}
+					for _, item := range contentResult.Array() {
+						typeStr := item.Get("type").String()
+						switch typeStr {
+						case "text":
+							flushThinking()
+							textBuilder.WriteString(item.Get("text").String())
+						case "tool_calls":
+							flushThinking()
+							flushText()
+							toolCalls := item.Get("tool_calls")
+							if toolCalls.IsArray() {
+								toolCalls.ForEach(func(_, tc gjson.Result) bool {
+									hasToolCall = true
+									toolUse := map[string]interface{}{
+										"type": "tool_use",
+										"id":   tc.Get("id").String(),
+										"name": tc.Get("function.name").String(),
+									}
 
-								argsStr := util.FixJSON(tc.Get("function.arguments").String())
-								if argsStr != "" {
-									var parsed interface{}
-									if err := json.Unmarshal([]byte(argsStr), &parsed); err == nil {
-										toolUse["input"] = parsed
+									argsStr := util.FixJSON(tc.Get("function.arguments").String())
+									if argsStr != "" {
+										var parsed interface{}
+										if err := json.Unmarshal([]byte(argsStr), &parsed); err == nil {
+											toolUse["input"] = parsed
+										} else {
+											toolUse["input"] = map[string]interface{}{}
+										}
 									} else {
 										toolUse["input"] = map[string]interface{}{}
 									}
-								} else {
-									toolUse["input"] = map[string]interface{}{}
-								}
 
-								contentBlocks = append(contentBlocks, toolUse)
-								return true
-							})
+									contentBlocks = append(contentBlocks, toolUse)
+									return true
+								})
+							}
+						case "reasoning":
+							flushText()
+							if thinking := item.Get("text"); thinking.Exists() {
+								thinkingBuilder.WriteString(thinking.String())
+							}
+						default:
+							flushThinking()
+							flushText()
 						}
-					case "reasoning":
-						flushText()
-						if thinking := item.Get("text"); thinking.Exists() {
-							thinkingBuilder.WriteString(thinking.String())
-						}
-					default:
-						flushThinking()
-						flushText()
+					}
+
+					flushThinking()
+					flushText()
+				} else if contentResult.Type == gjson.String {
+					textContent := contentResult.String()
+					if textContent != "" {
+						contentBlocks = append(contentBlocks, map[string]interface{}{
+							"type": "text",
+							"text": textContent,
+						})
 					}
 				}
-
-				flushThinking()
-				flushText()
 			}
 
 			if toolCalls := message.Get("tool_calls"); toolCalls.Exists() && toolCalls.IsArray() {

internal/translator/openai/gemini/openai_gemini_response.go

@@ -141,35 +141,46 @@ func ConvertOpenAIResponseToGemini(_ context.Context, _ string, originalRequestR
 					toolIndex := int(toolCall.Get("index").Int())
 					toolID := toolCall.Get("id").String()
 					toolType := toolCall.Get("type").String()
+					function := toolCall.Get("function")
 
-					if toolType == "function" {
-						function := toolCall.Get("function")
-						functionName := function.Get("name").String()
-						functionArgs := function.Get("arguments").String()
+					// Skip non-function tool calls explicitly marked as other types.
+					if toolType != "" && toolType != "function" {
+						return true
+					}
 
-						// Initialize accumulator if needed
-						if _, exists := (*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex]; !exists {
-							(*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex] = &ToolCallAccumulator{
-								ID:   toolID,
-								Name: functionName,
-							}
-						}
+					// OpenAI streaming deltas may omit the type field while still carrying function data.
+					if !function.Exists() {
+						return true
+					}
 
-						// Update ID if provided
-						if toolID != "" {
-							(*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex].ID = toolID
-						}
+					functionName := function.Get("name").String()
+					functionArgs := function.Get("arguments").String()
 
-						// Update name if provided
-						if functionName != "" {
-							(*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex].Name = functionName
-						}
-
-						// Accumulate arguments
-						if functionArgs != "" {
-							(*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex].Arguments.WriteString(functionArgs)
+					// Initialize accumulator if needed so later deltas without type can append arguments.
+					if _, exists := (*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex]; !exists {
+						(*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex] = &ToolCallAccumulator{
+							ID:   toolID,
+							Name: functionName,
 						}
 					}
+
+					acc := (*param).(*ConvertOpenAIResponseToGeminiParams).ToolCallsAccumulator[toolIndex]
+
+					// Update ID if provided
+					if toolID != "" {
+						acc.ID = toolID
+					}
+
+					// Update name if provided
+					if functionName != "" {
+						acc.Name = functionName
+					}
+
+					// Accumulate arguments
+					if functionArgs != "" {
+						acc.Arguments.WriteString(functionArgs)
+					}
+
 					return true
 				})
 

internal/usage/logger_plugin.go

@@ -91,6 +91,7 @@ type RequestDetail struct {
 	Timestamp time.Time  `json:"timestamp"`
 	Source    string     `json:"source"`
 	Tokens    TokenStats `json:"tokens"`
+	Failed    bool       `json:"failed"`
 }
 
 // TokenStats captures the token usage breakdown for a request.
@@ -165,7 +166,11 @@ func (s *RequestStatistics) Record(ctx context.Context, record coreusage.Record)
 	if statsKey == "" {
 		statsKey = resolveAPIIdentifier(ctx, record)
 	}
-	success := resolveSuccess(ctx)
+	failed := record.Failed
+	if !failed {
+		failed = !resolveSuccess(ctx)
+	}
+	success := !failed
 	modelName := record.Model
 	if modelName == "" {
 		modelName = "unknown"
@@ -193,6 +198,7 @@ func (s *RequestStatistics) Record(ctx context.Context, record coreusage.Record)
 		Timestamp: timestamp,
 		Source:    record.Source,
 		Tokens:    detail,
+		Failed:    failed,
 	})
 
 	s.requestsByDay[dayKey]++

internal/util/provider.go

@@ -4,6 +4,8 @@
 package util
 
 import (
+	"strings"
+
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/config"
 	"github.com/router-for-me/CLIProxyAPI/v6/internal/registry"
 )
@@ -141,3 +143,48 @@ func HideAPIKey(apiKey string) string {
 	}
 	return apiKey
 }
+
+// maskAuthorizationHeader masks the Authorization header value while preserving the auth type prefix.
+// Common formats: "Bearer <token>", "Basic <credentials>", "ApiKey <key>", etc.
+// It preserves the prefix (e.g., "Bearer ") and only masks the token/credential part.
+//
+// Parameters:
+//   - value: The Authorization header value
+//
+// Returns:
+//   - string: The masked Authorization value with prefix preserved
+func MaskAuthorizationHeader(value string) string {
+	parts := strings.SplitN(strings.TrimSpace(value), " ", 2)
+	if len(parts) < 2 {
+		return HideAPIKey(value)
+	}
+	return parts[0] + " " + HideAPIKey(parts[1])
+}
+
+// MaskSensitiveHeaderValue masks sensitive header values while preserving expected formats.
+//
+// Behavior by header key (case-insensitive):
+//   - "Authorization": Preserve the auth type prefix (e.g., "Bearer ") and mask only the credential part.
+//   - Headers containing "api-key": Mask the entire value using HideAPIKey.
+//   - Others: Return the original value unchanged.
+//
+// Parameters:
+//   - key:   The HTTP header name to inspect (case-insensitive matching).
+//   - value: The header value to mask when sensitive.
+//
+// Returns:
+//   - string: The masked value according to the header type; unchanged if not sensitive.
+func MaskSensitiveHeaderValue(key, value string) string {
+	lowerKey := strings.ToLower(strings.TrimSpace(key))
+	switch {
+	case lowerKey == "authorization":
+		return MaskAuthorizationHeader(value)
+	case strings.Contains(lowerKey, "api-key"),
+		strings.Contains(lowerKey, "apikey"),
+		strings.Contains(lowerKey, "token"),
+		strings.Contains(lowerKey, "secret"):
+		return HideAPIKey(value)
+	default:
+		return value
+	}
+}

internal/util/translator.go

@@ -212,161 +212,3 @@ func FixJSON(input string) string {
 
 	return out.String()
 }
-
-// SanitizeSchemaForGemini removes JSON Schema fields that are incompatible with Gemini API
-// to prevent "Proto field is not repeating, cannot start list" errors.
-//
-// Parameters:
-//   - schemaJSON: The JSON schema string to sanitize
-//
-// Returns:
-//   - string: The sanitized schema string
-//   - error: An error if the operation fails
-//
-// This function removes the following incompatible fields:
-// - additionalProperties: Not supported in Gemini function declarations
-// - $schema: JSON Schema meta-schema identifier, not needed for API
-// - allOf/anyOf/oneOf: Union type constructs not supported
-// - exclusiveMinimum/exclusiveMaximum: Advanced validation constraints
-// - patternProperties: Advanced property pattern matching
-// - dependencies: Property dependencies not supported
-// - type arrays: Converts ["string", "null"] to just "string"
-func SanitizeSchemaForGemini(schemaJSON string) (string, error) {
-	// Remove top-level incompatible fields
-	fieldsToRemove := []string{
-		"additionalProperties",
-		"$schema",
-		"allOf",
-		"anyOf",
-		"oneOf",
-		"exclusiveMinimum",
-		"exclusiveMaximum",
-		"patternProperties",
-		"dependencies",
-	}
-
-	result := schemaJSON
-	var err error
-
-	for _, field := range fieldsToRemove {
-		result, err = sjson.Delete(result, field)
-		if err != nil {
-			continue // Continue even if deletion fails
-		}
-	}
-
-	// Handle type arrays by converting them to single types
-	result = sanitizeTypeFields(result)
-
-	// Recursively clean nested objects
-	result = cleanNestedSchemas(result)
-
-	return result, nil
-}
-
-// sanitizeTypeFields converts type arrays to single types for Gemini compatibility
-func sanitizeTypeFields(jsonStr string) string {
-	// Parse the JSON to find all "type" fields
-	parsed := gjson.Parse(jsonStr)
-	result := jsonStr
-
-	// Walk through all paths to find type fields
-	var typeFields []string
-	walkForTypeFields(parsed, "", &typeFields)
-
-	// Process each type field
-	for _, path := range typeFields {
-		typeValue := gjson.Get(result, path)
-		if typeValue.IsArray() {
-			// Convert array to single type (prioritize string, then others)
-			arr := typeValue.Array()
-			if len(arr) > 0 {
-				var preferredType string
-				for _, t := range arr {
-					typeStr := t.String()
-					if typeStr == "string" {
-						preferredType = "string"
-						break
-					} else if typeStr == "number" || typeStr == "integer" {
-						preferredType = typeStr
-					} else if preferredType == "" {
-						preferredType = typeStr
-					}
-				}
-				if preferredType != "" {
-					result, _ = sjson.Set(result, path, preferredType)
-				}
-			}
-		}
-	}
-
-	return result
-}
-
-// walkForTypeFields recursively finds all "type" field paths in the JSON
-func walkForTypeFields(value gjson.Result, path string, paths *[]string) {
-	switch value.Type {
-	case gjson.JSON:
-		value.ForEach(func(key, val gjson.Result) bool {
-			var childPath string
-			if path == "" {
-				childPath = key.String()
-			} else {
-				childPath = path + "." + key.String()
-			}
-			if key.String() == "type" {
-				*paths = append(*paths, childPath)
-			}
-			walkForTypeFields(val, childPath, paths)
-			return true
-		})
-	default:
-
-	}
-}
-
-// cleanNestedSchemas recursively removes incompatible fields from nested schema objects
-func cleanNestedSchemas(jsonStr string) string {
-	fieldsToRemove := []string{"allOf", "anyOf", "oneOf", "exclusiveMinimum", "exclusiveMaximum"}
-
-	// Find all nested paths that might contain these fields
-	var pathsToClean []string
-	parsed := gjson.Parse(jsonStr)
-	findNestedSchemaPaths(parsed, "", fieldsToRemove, &pathsToClean)
-
-	result := jsonStr
-	// Remove fields from all found paths
-	for _, path := range pathsToClean {
-		result, _ = sjson.Delete(result, path)
-	}
-
-	return result
-}
-
-// findNestedSchemaPaths recursively finds paths containing incompatible schema fields
-func findNestedSchemaPaths(value gjson.Result, path string, fieldsToFind []string, paths *[]string) {
-	switch value.Type {
-	case gjson.JSON:
-		value.ForEach(func(key, val gjson.Result) bool {
-			var childPath string
-			if path == "" {
-				childPath = key.String()
-			} else {
-				childPath = path + "." + key.String()
-			}
-
-			// Check if this key is one we want to remove
-			for _, field := range fieldsToFind {
-				if key.String() == field {
-					*paths = append(*paths, childPath)
-					break
-				}
-			}
-
-			findNestedSchemaPaths(val, childPath, fieldsToFind, paths)
-			return true
-		})
-	default:
-
-	}
-}

internal/util/util.go

@@ -84,3 +84,17 @@ func CountAuthFiles(authDir string) int {
 	}
 	return count
 }
+
+// WritablePath returns the cleaned WRITABLE_PATH environment variable when it is set.
+// It accepts both uppercase and lowercase variants for compatibility with existing conventions.
+func WritablePath() string {
+	for _, key := range []string{"WRITABLE_PATH", "writable_path"} {
+		if value, ok := os.LookupEnv(key); ok {
+			trimmed := strings.TrimSpace(value)
+			if trimmed != "" {
+				return filepath.Clean(trimmed)
+			}
+		}
+	}
+	return ""
+}

internal/watcher/watcher.go

@@ -1192,6 +1192,9 @@ func buildConfigChangeDetails(oldCfg, newCfg *config.Config) []string {
 	if oldCfg.UsageStatisticsEnabled != newCfg.UsageStatisticsEnabled {
 		changes = append(changes, fmt.Sprintf("usage-statistics-enabled: %t -> %t", oldCfg.UsageStatisticsEnabled, newCfg.UsageStatisticsEnabled))
 	}
+	if oldCfg.DisableCooling != newCfg.DisableCooling {
+		changes = append(changes, fmt.Sprintf("disable-cooling: %t -> %t", oldCfg.DisableCooling, newCfg.DisableCooling))
+	}
 	if oldCfg.RequestLog != newCfg.RequestLog {
 		changes = append(changes, fmt.Sprintf("request-log: %t -> %t", oldCfg.RequestLog, newCfg.RequestLog))
 	}

sdk/auth/iflow.go

@@ -26,7 +26,7 @@ func (a *IFlowAuthenticator) Provider() string { return "iflow" }
 
 // RefreshLead indicates how soon before expiry a refresh should be attempted.
 func (a *IFlowAuthenticator) RefreshLead() *time.Duration {
-	d := 3 * time.Hour
+	d := 24 * time.Hour
 	return &d
 }
 

sdk/cliproxy/auth/manager.go

@@ -8,6 +8,7 @@ import (
 	"strconv"
 	"strings"
 	"sync"
+	"sync/atomic"
 	"time"
 
 	"github.com/google/uuid"
@@ -40,8 +41,17 @@ const (
 	refreshCheckInterval  = 5 * time.Second
 	refreshPendingBackoff = time.Minute
 	refreshFailureBackoff = 5 * time.Minute
+	quotaBackoffBase      = time.Second
+	quotaBackoffMax       = 30 * time.Minute
 )
 
+var quotaCooldownDisabled atomic.Bool
+
+// SetQuotaCooldownDisabled toggles quota cooldown scheduling globally.
+func SetQuotaCooldownDisabled(disable bool) {
+	quotaCooldownDisabled.Store(disable)
+}
+
 // Result captures execution outcome used to adjust auth state.
 type Result struct {
 	// AuthID references the auth that produced this result.
@@ -532,9 +542,18 @@ func (m *Manager) MarkResult(ctx context.Context, result Result) {
 					suspendReason = "payment_required"
 					shouldSuspendModel = true
 				case 429:
-					next := now.Add(30 * time.Minute)
+					cooldown, nextLevel := nextQuotaCooldown(state.Quota.BackoffLevel)
+					var next time.Time
+					if cooldown > 0 {
+						next = now.Add(cooldown)
+					}
 					state.NextRetryAfter = next
-					state.Quota = QuotaState{Exceeded: true, Reason: "quota", NextRecoverAt: next}
+					state.Quota = QuotaState{
+						Exceeded:      true,
+						Reason:        "quota",
+						NextRecoverAt: next,
+						BackoffLevel:  nextLevel,
+					}
 					suspendReason = "quota"
 					shouldSuspendModel = true
 					setModelQuota = true
@@ -608,6 +627,7 @@ func updateAggregatedAvailability(auth *Auth, now time.Time) {
 	earliestRetry := time.Time{}
 	quotaExceeded := false
 	quotaRecover := time.Time{}
+	maxBackoffLevel := 0
 	for _, state := range auth.ModelStates {
 		if state == nil {
 			continue
@@ -636,6 +656,9 @@ func updateAggregatedAvailability(auth *Auth, now time.Time) {
 			if quotaRecover.IsZero() || (!state.Quota.NextRecoverAt.IsZero() && state.Quota.NextRecoverAt.Before(quotaRecover)) {
 				quotaRecover = state.Quota.NextRecoverAt
 			}
+			if state.Quota.BackoffLevel > maxBackoffLevel {
+				maxBackoffLevel = state.Quota.BackoffLevel
+			}
 		}
 	}
 	auth.Unavailable = allUnavailable
@@ -648,10 +671,12 @@ func updateAggregatedAvailability(auth *Auth, now time.Time) {
 		auth.Quota.Exceeded = true
 		auth.Quota.Reason = "quota"
 		auth.Quota.NextRecoverAt = quotaRecover
+		auth.Quota.BackoffLevel = maxBackoffLevel
 	} else {
 		auth.Quota.Exceeded = false
 		auth.Quota.Reason = ""
 		auth.Quota.NextRecoverAt = time.Time{}
+		auth.Quota.BackoffLevel = 0
 	}
 }
 
@@ -685,6 +710,7 @@ func clearAuthStateOnSuccess(auth *Auth, now time.Time) {
 	auth.Quota.Exceeded = false
 	auth.Quota.Reason = ""
 	auth.Quota.NextRecoverAt = time.Time{}
+	auth.Quota.BackoffLevel = 0
 	auth.LastError = nil
 	auth.NextRetryAfter = time.Time{}
 	auth.UpdatedAt = now
@@ -734,8 +760,14 @@ func applyAuthFailureState(auth *Auth, resultErr *Error, now time.Time) {
 		auth.StatusMessage = "quota exhausted"
 		auth.Quota.Exceeded = true
 		auth.Quota.Reason = "quota"
-		auth.Quota.NextRecoverAt = now.Add(30 * time.Minute)
-		auth.NextRetryAfter = auth.Quota.NextRecoverAt
+		cooldown, nextLevel := nextQuotaCooldown(auth.Quota.BackoffLevel)
+		var next time.Time
+		if cooldown > 0 {
+			next = now.Add(cooldown)
+		}
+		auth.Quota.NextRecoverAt = next
+		auth.Quota.BackoffLevel = nextLevel
+		auth.NextRetryAfter = next
 	case 408, 500, 502, 503, 504:
 		auth.StatusMessage = "transient upstream error"
 		auth.NextRetryAfter = now.Add(1 * time.Minute)
@@ -746,6 +778,24 @@ func applyAuthFailureState(auth *Auth, resultErr *Error, now time.Time) {
 	}
 }
 
+// nextQuotaCooldown returns the next cooldown duration and updated backoff level for repeated quota errors.
+func nextQuotaCooldown(prevLevel int) (time.Duration, int) {
+	if prevLevel < 0 {
+		prevLevel = 0
+	}
+	if quotaCooldownDisabled.Load() {
+		return 0, prevLevel
+	}
+	cooldown := quotaBackoffBase * time.Duration(1<<prevLevel)
+	if cooldown < quotaBackoffBase {
+		cooldown = quotaBackoffBase
+	}
+	if cooldown >= quotaBackoffMax {
+		return quotaBackoffMax, prevLevel
+	}
+	return cooldown, prevLevel + 1
+}
+
 // List returns all auth entries currently known by the manager.
 func (m *Manager) List() []*Auth {
 	m.mu.RLock()

sdk/cliproxy/auth/types.go

@@ -65,6 +65,8 @@ type QuotaState struct {
 	Reason string `json:"reason,omitempty"`
 	// NextRecoverAt is when the credential may become available again.
 	NextRecoverAt time.Time `json:"next_recover_at"`
+	// BackoffLevel stores the progressive cooldown exponent used for rate limits.
+	BackoffLevel int `json:"backoff_level,omitempty"`
 }
 
 // ModelState captures the execution state for a specific model under an auth entry.

sdk/cliproxy/usage/manager.go

@@ -16,6 +16,7 @@ type Record struct {
 	AuthID      string
 	Source      string
 	RequestedAt time.Time
+	Failed      bool
 	Detail      Detail
 }