fix: chat with screenshot

internal/version/VERSION

@@ -1 +1 @@
-v5.0.0-beta-2505211805
+v5.0.0-beta-2505212235

mcphost/chat.go

@@ -16,6 +16,7 @@ import (
 	"github.com/cloudwego/eino/schema"
 	"github.com/httprunner/httprunner/v5/uixt/ai"
 	"github.com/httprunner/httprunner/v5/uixt/option"
+	"github.com/mark3labs/mcp-go/mcp"
 	"github.com/pkg/errors"
 	"github.com/rs/zerolog/log"
 	"golang.org/x/term"
@@ -174,24 +175,46 @@ func (c *Chat) handleToolCalls(ctx context.Context, toolCalls []schema.ToolCall)
 			continue
 		}
 
-		// Format tool result
-		resultStr := ""
+		// Format tool result, append message to history
+		renderStr := ""
 		if result != nil && len(result.Content) > 0 {
 			for _, item := range result.Content {
-				resultStr += fmt.Sprintf("%v\n", item)
+				if contentMap, ok := item.(mcp.TextContent); ok {
+					renderStr += contentMap.Text + "\n"
+					toolMsg := &schema.Message{
+						Role:       schema.Tool,
+						ToolCallID: toolCall.ID,
+						Content:    contentMap.Text,
+					}
+					c.planner.History().Append(toolMsg)
+				} else if contentMap, ok := item.(mcp.ImageContent); ok {
+					renderStr += "<data:image/base64...>\n" // base64-encoded image data
+					toolMsg := &schema.Message{
+						Role:       schema.Tool,
+						ToolCallID: toolCall.ID,
+						MultiContent: []schema.ChatMessagePart{
+							{
+								Type: schema.ChatMessagePartTypeImageURL,
+								ImageURL: &schema.ChatMessageImageURL{
+									URL:      contentMap.Data,
+									MIMEType: contentMap.MIMEType,
+								},
+							},
+						},
+					}
+					c.planner.History().Append(toolMsg)
+				}
 			}
 		} else {
-			resultStr = fmt.Sprintf("%+v", result)
+			renderStr = fmt.Sprintf("%+v", result)
+			toolMsg := &schema.Message{
+				Role:       schema.Tool,
+				ToolCallID: toolCall.ID,
+				Content:    renderStr,
+			}
+			c.planner.History().Append(toolMsg)
 		}
-		c.renderContent("Tool Result", resultStr)
-
-		// Add tool result to history
-		toolMsg := &schema.Message{
-			Role:       schema.Tool,
-			Content:    resultStr,
-			ToolCallID: toolCall.ID,
-		}
-		c.planner.History().Append(toolMsg)
+		c.renderContent("Tool Result", renderStr)
 	}
 	return nil
 }

uixt/ai/ai.go

@@ -95,14 +95,12 @@ func GetModelConfig(modelType option.LLMServiceType) (*ModelConfig, error) {
 			"env %s missed", EnvModelName)
 	}
 
-	maxTokens := 4096
 	temperature := float32(0.7)
 	modelConfig := &openai.ChatModelConfig{
 		BaseURL:     openaiBaseURL,
 		APIKey:      openaiAPIKey,
 		Model:       modelName,
 		Timeout:     defaultTimeout,
-		MaxTokens:   &maxTokens,
 		Temperature: &temperature,
 	}
 

uixt/ai/session.go

@@ -100,6 +100,15 @@ func logRequest(messages ConversationHistory) {
 func logResponse(message *schema.Message) {
 	logger := log.Info().Str("role", string(message.Role)).
 		Str("content", message.Content)
+
+	var toolCalls []string
+	if len(message.ToolCalls) > 0 {
+		for _, toolCall := range message.ToolCalls {
+			toolCalls = append(toolCalls, toolCall.Function.Name)
+		}
+		logger = logger.Strs("tool_calls", toolCalls)
+	}
+
 	if message.ResponseMeta != nil {
 		logger = logger.Str("finish_reason", message.ResponseMeta.FinishReason)
 		// Log usage statistics

uixt/driver_ext_screenshot.go

@@ -292,6 +292,7 @@ func saveScreenShot(raw *bytes.Buffer, screenshotPath string) error {
 }
 
 func compressImageBuffer(raw *bytes.Buffer) (compressed *bytes.Buffer, err error) {
+	rawSize := raw.Len()
 	// decode image from buffer
 	img, format, err := image.Decode(raw)
 	if err != nil {
@@ -312,6 +313,10 @@ func compressImageBuffer(raw *bytes.Buffer) (compressed *bytes.Buffer, err error
 		return nil, fmt.Errorf("unsupported image format: %s", format)
 	}
 
+	compressedSize := buf.Len()
+	log.Debug().Int("rawSize", rawSize).Int("compressedSize", compressedSize).
+		Msg("compress image buffer")
+
 	// return compressed image buffer
 	return &buf, nil
 }