Refactor role handling in OpenAI request processing for improved clarity and functionality

- Simplified variable initialization in `browser.go` for readability.
- Updated error handling in `request_logger.go` with better resource cleanup using deferred anonymous functions.

Refactor API handlers to use `GetContextWithCancel` for streamlined context creation and response handling

- Replaced redundant `context.WithCancel` and `context.WithValue` logic with the new `GetContextWithCancel` utility in all handlers.
- Centralized API response storage in the given context during cancellation.
- Updated associated cancellation calls for consistency and improved resource management.

- Replaced `apiResponseData` with `AddAPIResponseData` for centralized response recording.
- Simplified cancellation logic by switching to a boolean-based `cliCancel` method.
- Removed unused `apiResponseData` slices across handlers to reduce memory usage.
- Updated `handlers.go` to support unified response data storage per request context.

internal/api/handlers/claude/code_handlers.go

@@ -108,8 +108,7 @@ func (h *ClaudeCodeAPIHandlers) handleGeminiStreamingResponse(c *gin.Context, ra
 
 	// Create a cancellable context for the backend client request
 	// This allows proper cleanup and cancellation of ongoing requests
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	cliClient = client.NewGeminiClient(nil, nil, nil)
@@ -159,7 +158,6 @@ outLoop:
 		responseType := 0
 		responseIndex := 0
 
-		apiResponseData := make([]byte, 0)
 		// Main streaming loop - handles multiple concurrent events using Go channels
 		// This select statement manages four different types of events simultaneously
 		for {
@@ -169,7 +167,6 @@ outLoop:
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("GeminiClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request to prevent resource leaks
 					return
 				}
@@ -186,12 +183,11 @@ outLoop:
 					_, _ = c.Writer.Write([]byte("\n\n\n"))
 
 					flusher.Flush()
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
 
-				apiResponseData = append(apiResponseData, chunk...)
+				h.AddAPIResponseData(c, chunk)
 				// Convert the backend response to Claude-compatible format
 				// This translation layer ensures API compatibility
 				claudeFormat := translatorClaudeCodeToGeminiCli.ConvertCliResponseToClaudeCode(chunk, isGlAPIKey, hasFirstResponse, &responseType, &responseIndex)
@@ -214,8 +210,7 @@ outLoop:
 						c.Status(errInfo.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, errInfo.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(errInfo.Error.Error()))
-						cliCancel()
+						cliCancel(errInfo.Error)
 					}
 					return
 				}
@@ -280,8 +275,7 @@ func (h *ClaudeCodeAPIHandlers) handleCodexStreamingResponse(c *gin.Context, raw
 	// return
 	// Create a cancellable context for the backend client request
 	// This allows proper cleanup and cancellation of ongoing requests
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -316,7 +310,6 @@ outLoop:
 		hasFirstResponse := false
 		hasToolCall := false
 
-		apiResponseData := make([]byte, 0)
 		// Main streaming loop - handles multiple concurrent events using Go channels
 		// This select statement manages four different types of events simultaneously
 		for {
@@ -326,7 +319,6 @@ outLoop:
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request to prevent resource leaks
 					return
 				}
@@ -336,11 +328,12 @@ outLoop:
 			case chunk, okStream := <-respChan:
 				if !okStream {
 					flusher.Flush()
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
+
 				// Convert the backend response to Claude-compatible format
 				// This translation layer ensures API compatibility
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
@@ -371,9 +364,8 @@ outLoop:
 						// Forward other errors directly to the client
 						c.Status(errInfo.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, errInfo.Error.Error())
-						c.Set("API_RESPONSE", []byte(errInfo.Error.Error()))
 						flusher.Flush()
-						cliCancel()
+						cliCancel(errInfo.Error)
 					}
 					return
 				}

internal/api/handlers/gemini/cli/cli_handlers.go

@@ -156,8 +156,7 @@ func (h *GeminiCLIAPIHandlers) handleInternalStreamGenerateContent(c *gin.Contex
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	modelName := modelResult.String()
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -188,26 +187,23 @@ outLoop:
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJSON, "")
 		hasFirstResponse := false
 
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("GeminiClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
 
-				apiResponseData = append(apiResponseData, chunk...)
+				h.AddAPIResponseData(c, chunk)
 
 				hasFirstResponse = true
 				if cliClient.(*client.GeminiClient).GetGenerativeLanguageAPIKey() != "" {
@@ -227,8 +223,7 @@ outLoop:
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}
@@ -248,8 +243,8 @@ func (h *GeminiCLIAPIHandlers) handleInternalGenerateContent(c *gin.Context, raw
 	// log.Debugf("GenerateContent: %s", string(rawJSON))
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	modelName := modelResult.String()
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -282,14 +277,12 @@ func (h *GeminiCLIAPIHandlers) handleInternalGenerateContent(c *gin.Context, raw
 				c.Status(err.StatusCode)
 				_, _ = c.Writer.Write([]byte(err.Error.Error()))
 				log.Debugf("code: %d, error: %s", err.StatusCode, err.Error.Error())
-				c.Set("API_RESPONSE", []byte(err.Error.Error()))
-				cliCancel()
+				cliCancel(err.Error)
 			}
 			break
 		} else {
 			_, _ = c.Writer.Write(resp)
-			c.Set("API_RESPONSE", resp)
-			cliCancel()
+			cliCancel(resp)
 			break
 		}
 	}
@@ -328,8 +321,7 @@ func (h *GeminiCLIAPIHandlers) handleCodexInternalStreamGenerateContent(c *gin.C
 
 	modelName := gjson.GetBytes(rawJSON, "model")
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -362,7 +354,6 @@ outLoop:
 			ResponseID:        "",
 			LastStorageOutput: "",
 		}
-		apiResponseData := make([]byte, 0)
 
 		for {
 			select {
@@ -370,20 +361,19 @@ outLoop:
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
 				// _, _ = logFile.Write(chunk)
 				// _, _ = logFile.Write([]byte("\n"))
-				apiResponseData = append(apiResponseData, chunk...)
+				h.AddAPIResponseData(c, chunk)
+
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
 					jsonData := chunk[6:]
 					data := gjson.ParseBytes(jsonData)
@@ -411,8 +401,7 @@ outLoop:
 						c.Status(errMessage.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, errMessage.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(errMessage.Error.Error()))
-						cliCancel()
+						cliCancel(errMessage.Error)
 					}
 					return
 				}
@@ -438,8 +427,7 @@ func (h *GeminiCLIAPIHandlers) handleCodexInternalGenerateContent(c *gin.Context
 
 	modelName := gjson.GetBytes(rawJSON, "model")
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -464,25 +452,24 @@ outLoop:
 
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, []byte(newRequestJSON), "")
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
+
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
 					jsonData := chunk[6:]
 					data := gjson.ParseBytes(jsonData)
@@ -506,8 +493,7 @@ outLoop:
 						log.Debugf("org: %s", string(orgRawJSON))
 						log.Debugf("raw: %s", string(rawJSON))
 						log.Debugf("newRequestJSON: %s", newRequestJSON)
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}

internal/api/handlers/gemini/gemini_handlers.go

@@ -42,20 +42,19 @@ func NewGeminiAPIHandlers(apiHandlers *handlers.APIHandlers) *GeminiAPIHandlers
 // It returns a JSON response containing available Gemini models and their specifications.
 func (h *GeminiAPIHandlers) GeminiModels(c *gin.Context) {
 	c.JSON(http.StatusOK, gin.H{
-		"data": []map[string]any{
+		"models": []map[string]any{
 			{
-				"id":                    "gemini-2.5-flash",
-				"object":                "model",
-				"version":               "001",
-				"name":                  "Gemini 2.5 Flash",
-				"description":           "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
-				"context_length":        1_048_576,
-				"max_completion_tokens": 65_536,
-				"supported_parameters": []string{
-					"tools",
-					"temperature",
-					"top_p",
-					"top_k",
+				"name":             "models/gemini-2.5-flash",
+				"version":          "001",
+				"displayName":      "Gemini 2.5 Flash",
+				"description":      "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
+				"inputTokenLimit":  1048576,
+				"outputTokenLimit": 65536,
+				"supportedGenerationMethods": []string{
+					"generateContent",
+					"countTokens",
+					"createCachedContent",
+					"batchGenerateContent",
 				},
 				"temperature":    1,
 				"topP":           0.95,
@@ -64,18 +63,17 @@ func (h *GeminiAPIHandlers) GeminiModels(c *gin.Context) {
 				"thinking":       true,
 			},
 			{
-				"id":                    "gemini-2.5-pro",
-				"object":                "model",
-				"version":               "2.5",
-				"name":                  "Gemini 2.5 Pro",
-				"description":           "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
-				"context_length":        1_048_576,
-				"max_completion_tokens": 65_536,
-				"supported_parameters": []string{
-					"tools",
-					"temperature",
-					"top_p",
-					"top_k",
+				"name":             "models/gemini-2.5-pro",
+				"version":          "2.5",
+				"displayName":      "Gemini 2.5 Pro",
+				"description":      "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
+				"inputTokenLimit":  1048576,
+				"outputTokenLimit": 65536,
+				"supportedGenerationMethods": []string{
+					"generateContent",
+					"countTokens",
+					"createCachedContent",
+					"batchGenerateContent",
 				},
 				"temperature":    1,
 				"topP":           0.95,
@@ -84,15 +82,14 @@ func (h *GeminiAPIHandlers) GeminiModels(c *gin.Context) {
 				"thinking":       true,
 			},
 			{
-				"id":                    "gpt-5",
-				"object":                "model",
-				"version":               "gpt-5-2025-08-07",
-				"name":                  "GPT 5",
-				"description":           "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
-				"context_length":        400_000,
-				"max_completion_tokens": 128_000,
-				"supported_parameters": []string{
-					"tools",
+				"name":             "gpt-5",
+				"version":          "001",
+				"displayName":      "GPT 5",
+				"description":      "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+				"inputTokenLimit":  400000,
+				"outputTokenLimit": 128000,
+				"supportedGenerationMethods": []string{
+					"generateContent",
 				},
 				"temperature":    1,
 				"topP":           0.95,
@@ -122,39 +119,38 @@ func (h *GeminiAPIHandlers) GeminiGetHandler(c *gin.Context) {
 	switch request.Action {
 	case "gemini-2.5-pro":
 		c.JSON(http.StatusOK, gin.H{
-			"id":                    "gemini-2.5-pro",
-			"object":                "model",
-			"version":               "2.5",
-			"name":                  "Gemini 2.5 Pro",
-			"description":           "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
-			"context_length":        1_048_576,
-			"max_completion_tokens": 65_536,
-			"supported_parameters": []string{
-				"tools",
-				"temperature",
-				"top_p",
-				"top_k",
+			"name":             "models/gemini-2.5-pro",
+			"version":          "2.5",
+			"displayName":      "Gemini 2.5 Pro",
+			"description":      "Stable release (June 17th, 2025) of Gemini 2.5 Pro",
+			"inputTokenLimit":  1048576,
+			"outputTokenLimit": 65536,
+			"supportedGenerationMethods": []string{
+				"generateContent",
+				"countTokens",
+				"createCachedContent",
+				"batchGenerateContent",
 			},
 			"temperature":    1,
 			"topP":           0.95,
 			"topK":           64,
 			"maxTemperature": 2,
 			"thinking":       true,
-		})
+		},
+		)
 	case "gemini-2.5-flash":
 		c.JSON(http.StatusOK, gin.H{
-			"id":                    "gemini-2.5-flash",
-			"object":                "model",
-			"version":               "001",
-			"name":                  "Gemini 2.5 Flash",
-			"description":           "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
-			"context_length":        1_048_576,
-			"max_completion_tokens": 65_536,
-			"supported_parameters": []string{
-				"tools",
-				"temperature",
-				"top_p",
-				"top_k",
+			"name":             "models/gemini-2.5-flash",
+			"version":          "001",
+			"displayName":      "Gemini 2.5 Flash",
+			"description":      "Stable version of Gemini 2.5 Flash, our mid-size multimodal model that supports up to 1 million tokens, released in June of 2025.",
+			"inputTokenLimit":  1048576,
+			"outputTokenLimit": 65536,
+			"supportedGenerationMethods": []string{
+				"generateContent",
+				"countTokens",
+				"createCachedContent",
+				"batchGenerateContent",
 			},
 			"temperature":    1,
 			"topP":           0.95,
@@ -164,15 +160,14 @@ func (h *GeminiAPIHandlers) GeminiGetHandler(c *gin.Context) {
 		})
 	case "gpt-5":
 		c.JSON(http.StatusOK, gin.H{
-			"id":                    "gpt-5",
-			"object":                "model",
-			"version":               "gpt-5-2025-08-07",
-			"name":                  "GPT 5",
-			"description":           "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
-			"context_length":        400_000,
-			"max_completion_tokens": 128_000,
-			"supported_parameters": []string{
-				"tools",
+			"name":             "gpt-5",
+			"version":          "001",
+			"displayName":      "GPT 5",
+			"description":      "Stable version of GPT 5, The best model for coding and agentic tasks across domains.",
+			"inputTokenLimit":  400000,
+			"outputTokenLimit": 128000,
+			"supportedGenerationMethods": []string{
+				"generateContent",
 			},
 			"temperature":    1,
 			"topP":           0.95,
@@ -267,8 +262,7 @@ func (h *GeminiAPIHandlers) handleGeminiStreamGenerateContent(c *gin.Context, ra
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	modelName := modelResult.String()
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -329,25 +323,23 @@ outLoop:
 
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, rawJSON, alt)
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("GeminiClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
 
 				if cliClient.(*client.GeminiClient).GetGenerativeLanguageAPIKey() == "" {
 					if alt == "" {
@@ -389,8 +381,7 @@ outLoop:
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}
@@ -408,8 +399,7 @@ func (h *GeminiAPIHandlers) handleGeminiCountTokens(c *gin.Context, rawJSON []by
 	// orgrawJSON := rawJSON
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	modelName := modelResult.String()
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -451,8 +441,7 @@ func (h *GeminiAPIHandlers) handleGeminiCountTokens(c *gin.Context, rawJSON []by
 			} else {
 				c.Status(err.StatusCode)
 				_, _ = c.Writer.Write([]byte(err.Error.Error()))
-				c.Set("API_RESPONSE", []byte(err.Error.Error()))
-				cliCancel()
+				cliCancel(err.Error)
 				// log.Debugf(err.Error.Error())
 				// log.Debugf(string(rawJSON))
 				// log.Debugf(string(orgrawJSON))
@@ -466,8 +455,7 @@ func (h *GeminiAPIHandlers) handleGeminiCountTokens(c *gin.Context, rawJSON []by
 				}
 			}
 			_, _ = c.Writer.Write(resp)
-			c.Set("API_RESPONSE", resp)
-			cliCancel()
+			cliCancel(resp)
 			break
 		}
 	}
@@ -480,8 +468,7 @@ func (h *GeminiAPIHandlers) handleGeminiGenerateContent(c *gin.Context, rawJSON
 
 	modelResult := gjson.GetBytes(rawJSON, "model")
 	modelName := modelResult.String()
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -543,8 +530,7 @@ func (h *GeminiAPIHandlers) handleGeminiGenerateContent(c *gin.Context, rawJSON
 			} else {
 				c.Status(err.StatusCode)
 				_, _ = c.Writer.Write([]byte(err.Error.Error()))
-				c.Set("API_RESPONSE", []byte(err.Error.Error()))
-				cliCancel()
+				cliCancel(err.Error)
 			}
 			break
 		} else {
@@ -555,8 +541,7 @@ func (h *GeminiAPIHandlers) handleGeminiGenerateContent(c *gin.Context, rawJSON
 				}
 			}
 			_, _ = c.Writer.Write(resp)
-			c.Set("API_RESPONSE", resp)
-			cliCancel()
+			cliCancel(resp)
 			break
 		}
 	}
@@ -586,8 +571,7 @@ func (h *GeminiAPIHandlers) handleCodexStreamGenerateContent(c *gin.Context, raw
 
 	modelName := gjson.GetBytes(rawJSON, "model")
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -614,8 +598,6 @@ outLoop:
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, []byte(newRequestJSON), "")
 
-		apiResponseData := make([]byte, 0)
-
 		params := &translatorGeminiToCodex.ConvertCodexResponseToGeminiParams{
 			Model:             modelName.String(),
 			CreatedAt:         0,
@@ -628,18 +610,17 @@ outLoop:
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
 
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
 					jsonData := chunk[6:]
@@ -667,8 +648,7 @@ outLoop:
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}
@@ -688,8 +668,7 @@ func (h *GeminiAPIHandlers) handleCodexGenerateContent(c *gin.Context, rawJSON [
 
 	modelName := gjson.GetBytes(rawJSON, "model")
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -714,25 +693,23 @@ outLoop:
 
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, []byte(newRequestJSON), "")
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
 
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
 					jsonData := chunk[6:]
@@ -754,8 +731,7 @@ outLoop:
 					} else {
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}

internal/api/handlers/handlers.go

@@ -12,6 +12,7 @@ import (
 	"github.com/luispater/CLIProxyAPI/internal/config"
 	"github.com/luispater/CLIProxyAPI/internal/util"
 	log "github.com/sirupsen/logrus"
+	"golang.org/x/net/context"
 )
 
 // ErrorResponse represents a standard error response format for the API.
@@ -50,6 +51,9 @@ type APIHandlers struct {
 	// LastUsedClientIndex tracks the last used client index for each provider
 	// to implement round-robin load balancing.
 	LastUsedClientIndex map[string]int
+
+	// apiResponseData recording provider api response data
+	apiResponseData map[*gin.Context][]byte
 }
 
 // NewAPIHandlers creates a new API handlers instance.
@@ -67,6 +71,7 @@ func NewAPIHandlers(cliClients []client.Client, cfg *config.Config) *APIHandlers
 		Cfg:                 cfg,
 		Mutex:               &sync.Mutex{},
 		LastUsedClientIndex: make(map[string]int),
+		apiResponseData:     make(map[*gin.Context][]byte),
 	}
 }
 
@@ -185,3 +190,43 @@ func (h *APIHandlers) GetAlt(c *gin.Context) string {
 	}
 	return alt
 }
+
+func (h *APIHandlers) GetContextWithCancel(c *gin.Context, ctx context.Context) (context.Context, APIHandlerCancelFunc) {
+	newCtx, cancel := context.WithCancel(ctx)
+	newCtx = context.WithValue(newCtx, "gin", c)
+	return newCtx, func(params ...interface{}) {
+		if h.Cfg.RequestLog {
+			if len(params) == 1 {
+				data := params[0]
+				switch data.(type) {
+				case []byte:
+					c.Set("API_RESPONSE", data.([]byte))
+				case error:
+					c.Set("API_RESPONSE", []byte(data.(error).Error()))
+				case string:
+					c.Set("API_RESPONSE", []byte(data.(string)))
+				case bool:
+				case nil:
+				}
+			} else {
+				if _, hasKey := h.apiResponseData[c]; hasKey {
+					c.Set("API_RESPONSE", h.apiResponseData[c])
+					delete(h.apiResponseData, c)
+				}
+			}
+		}
+
+		cancel()
+	}
+}
+
+func (h *APIHandlers) AddAPIResponseData(c *gin.Context, data []byte) {
+	if h.Cfg.RequestLog {
+		if _, hasKey := h.apiResponseData[c]; !hasKey {
+			h.apiResponseData[c] = make([]byte, 0)
+		}
+		h.apiResponseData[c] = append(h.apiResponseData[c], data...)
+	}
+}
+
+type APIHandlerCancelFunc func(params ...interface{})

internal/api/handlers/openai/openai_handlers.go

@@ -160,8 +160,7 @@ func (h *OpenAIAPIHandlers) handleGeminiNonStreamingResponse(c *gin.Context, raw
 	c.Header("Content-Type", "application/json")
 
 	modelName, systemInstruction, contents, tools := translatorOpenAIToGeminiCli.ConvertOpenAIChatRequestToCli(rawJSON)
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -195,8 +194,7 @@ func (h *OpenAIAPIHandlers) handleGeminiNonStreamingResponse(c *gin.Context, raw
 			} else {
 				c.Status(err.StatusCode)
 				_, _ = c.Writer.Write([]byte(err.Error.Error()))
-				c.Set("API_RESPONSE", []byte(err.Error.Error()))
-				cliCancel()
+				cliCancel(err.Error)
 			}
 			break
 		} else {
@@ -204,8 +202,7 @@ func (h *OpenAIAPIHandlers) handleGeminiNonStreamingResponse(c *gin.Context, raw
 			if openAIFormat != "" {
 				_, _ = c.Writer.Write([]byte(openAIFormat))
 			}
-			c.Set("API_RESPONSE", resp)
-			cliCancel()
+			cliCancel(resp)
 			break
 		}
 	}
@@ -238,8 +235,7 @@ func (h *OpenAIAPIHandlers) handleGeminiStreamingResponse(c *gin.Context, rawJSO
 
 	// Prepare the request for the backend client.
 	modelName, systemInstruction, contents, tools := translatorOpenAIToGeminiCli.ConvertOpenAIChatRequestToCli(rawJSON)
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -270,7 +266,6 @@ outLoop:
 		}
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendMessageStream(cliCtx, rawJSON, modelName, systemInstruction, contents, tools)
-		apiResponseData := make([]byte, 0)
 
 		hasFirstResponse := false
 		for {
@@ -279,7 +274,6 @@ outLoop:
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("GeminiClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
@@ -289,11 +283,12 @@ outLoop:
 					// Stream is closed, send the final [DONE] message.
 					_, _ = fmt.Fprintf(c.Writer, "data: [DONE]\n\n")
 					flusher.Flush()
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
+
 				// Convert the chunk to OpenAI format and send it to the client.
 				hasFirstResponse = true
 				openAIFormat := translatorOpenAIToGeminiCli.ConvertCliResponseToOpenAIChat(chunk, time.Now().Unix(), isGlAPIKey)
@@ -310,8 +305,7 @@ outLoop:
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
 						flusher.Flush()
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}
@@ -338,8 +332,8 @@ func (h *OpenAIAPIHandlers) handleCodexNonStreamingResponse(c *gin.Context, rawJ
 
 	newRequestJSON := translatorOpenAIToCodex.ConvertOpenAIChatRequestToCodex(rawJSON)
 	modelName := gjson.GetBytes(rawJSON, "model")
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -363,25 +357,24 @@ outLoop:
 
 		// Send the message and receive response chunks and errors via channels.
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, []byte(newRequestJSON), "")
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
 			// Process incoming response chunks.
 			case chunk, okStream := <-respChan:
 				if !okStream {
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
+
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
 					jsonData := chunk[6:]
 					data := gjson.ParseBytes(jsonData)
@@ -400,8 +393,7 @@ outLoop:
 					} else {
 						c.Status(err.StatusCode)
 						_, _ = c.Writer.Write([]byte(err.Error.Error()))
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}
@@ -443,8 +435,7 @@ func (h *OpenAIAPIHandlers) handleCodexStreamingResponse(c *gin.Context, rawJSON
 
 	modelName := gjson.GetBytes(rawJSON, "model")
 
-	backgroundCtx, cliCancel := context.WithCancel(context.Background())
-	cliCtx := context.WithValue(backgroundCtx, "gin", c)
+	cliCtx, cliCancel := h.GetContextWithCancel(c, context.Background())
 
 	var cliClient client.Client
 	defer func() {
@@ -471,14 +462,12 @@ outLoop:
 		// Send the message and receive response chunks and errors via channels.
 		var params *translatorOpenAIToCodex.ConvertCliToOpenAIParams
 		respChan, errChan := cliClient.SendRawMessageStream(cliCtx, []byte(newRequestJSON), "")
-		apiResponseData := make([]byte, 0)
 		for {
 			select {
 			// Handle client disconnection.
 			case <-c.Request.Context().Done():
 				if c.Request.Context().Err().Error() == "context canceled" {
 					log.Debugf("CodexClient disconnected: %v", c.Request.Context().Err())
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel() // Cancel the backend request.
 					return
 				}
@@ -487,11 +476,12 @@ outLoop:
 				if !okStream {
 					_, _ = c.Writer.Write([]byte("[done]\n\n"))
 					flusher.Flush()
-					c.Set("API_RESPONSE", apiResponseData)
 					cliCancel()
 					return
 				}
-				apiResponseData = append(apiResponseData, chunk...)
+
+				h.AddAPIResponseData(c, chunk)
+
 				// log.Debugf("Response: %s\n", string(chunk))
 				// Convert the chunk to OpenAI format and send it to the client.
 				if bytes.HasPrefix(chunk, []byte("data: ")) {
@@ -518,9 +508,8 @@ outLoop:
 					} else {
 						c.Status(err.StatusCode)
 						_, _ = fmt.Fprint(c.Writer, err.Error.Error())
-						c.Set("API_RESPONSE", []byte(err.Error.Error()))
 						flusher.Flush()
-						cliCancel()
+						cliCancel(err.Error)
 					}
 					return
 				}

internal/browser/browser.go

@@ -105,7 +105,7 @@ func GetPlatformInfo() map[string]interface{} {
 		info["default_command"] = "rundll32"
 	case "linux":
 		browsers := []string{"xdg-open", "x-www-browser", "www-browser", "firefox", "chromium", "google-chrome"}
-		availableBrowsers := []string{}
+		var availableBrowsers []string
 		for _, browser := range browsers {
 			if _, err := exec.LookPath(browser); err == nil {
 				availableBrowsers = append(availableBrowsers, browser)

internal/logging/request_logger.go

@@ -85,7 +85,7 @@ func (l *FileRequestLogger) LogRequest(url, method string, requestHeaders map[st
 	content := l.formatLogContent(url, method, requestHeaders, body, apiRequest, apiResponse, decompressedResponse, statusCode, responseHeaders)
 
 	// Write to file
-	if err := os.WriteFile(filePath, []byte(content), 0644); err != nil {
+	if err = os.WriteFile(filePath, []byte(content), 0644); err != nil {
 		return fmt.Errorf("failed to write log file: %w", err)
 	}
 
@@ -115,7 +115,7 @@ func (l *FileRequestLogger) LogStreamingRequest(url, method string, headers map[
 
 	// Write initial request information
 	requestInfo := l.formatRequestInfo(url, method, headers, body)
-	if _, err := file.WriteString(requestInfo); err != nil {
+	if _, err = file.WriteString(requestInfo); err != nil {
 		_ = file.Close()
 		return nil, fmt.Errorf("failed to write request info: %w", err)
 	}
@@ -257,7 +257,9 @@ func (l *FileRequestLogger) decompressGzip(data []byte) ([]byte, error) {
 	if err != nil {
 		return nil, fmt.Errorf("failed to create gzip reader: %w", err)
 	}
-	defer reader.Close()
+	defer func() {
+		_ = reader.Close()
+	}()
 
 	decompressed, err := io.ReadAll(reader)
 	if err != nil {
@@ -270,7 +272,9 @@ func (l *FileRequestLogger) decompressGzip(data []byte) ([]byte, error) {
 // decompressDeflate decompresses deflate-encoded data.
 func (l *FileRequestLogger) decompressDeflate(data []byte) ([]byte, error) {
 	reader := flate.NewReader(bytes.NewReader(data))
-	defer reader.Close()
+	defer func() {
+		_ = reader.Close()
+	}()
 
 	decompressed, err := io.ReadAll(reader)
 	if err != nil {

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

@@ -94,9 +94,10 @@ func ConvertOpenAIChatRequestToCli(rawJSON []byte) (string, *client.Content, []c
 				continue
 			}
 
-			switch roleResult.String() {
-			// System messages are converted to a user message followed by a model's acknowledgment.
-			case "system":
+			role := roleResult.String()
+
+			if role == "system" && len(messagesResults) > 1 {
+				// System messages are converted to a user message followed by a model's acknowledgment.
 				if contentResult.Type == gjson.String {
 					systemInstruction = &client.Content{Role: "user", Parts: []client.Part{{Text: contentResult.String()}}}
 				} else if contentResult.IsObject() {
@@ -105,8 +106,8 @@ func ConvertOpenAIChatRequestToCli(rawJSON []byte) (string, *client.Content, []c
 						systemInstruction = &client.Content{Role: "user", Parts: []client.Part{{Text: contentResult.Get("text").String()}}}
 					}
 				}
-			// User messages can contain simple text or a multi-part body.
-			case "user":
+			} else if role == "user" || (role == "system" && len(messagesResults) == 1) { // If there's only a system message, treat it as a user message.
+				// User messages can contain simple text or a multi-part body.
 				if contentResult.Type == gjson.String {
 					contents = append(contents, client.Content{Role: "user", Parts: []client.Part{{Text: contentResult.String()}}})
 				} else if contentResult.IsArray() {
@@ -151,9 +152,10 @@ func ConvertOpenAIChatRequestToCli(rawJSON []byte) (string, *client.Content, []c
 					}
 					contents = append(contents, client.Content{Role: "user", Parts: parts})
 				}
-			// Assistant messages can contain text responses or tool calls
-			// In the internal format, assistant messages are converted to "model" role
-			case "assistant":
+			} else if role == "assistant" {
+				// Assistant messages can contain text responses or tool calls
+				// In the internal format, assistant messages are converted to "model" role
+
 				if contentResult.Type == gjson.String {
 					// Simple text response from the assistant
 					contents = append(contents, client.Content{Role: "model", Parts: []client.Part{{Text: contentResult.String()}}})