@xagent-ai/cli 1.3.0 → 1.3.2

package/src/conversation.ts

@@ -2,6 +2,7 @@ import fs from 'fs/promises';
 import path from 'path';
 import os from 'os';
 import { ChatMessage, Conversation } from './types.js';
+import { output as logOutput } from './output-util.js';
 
 export class ConversationManager {
   private conversationsDir: string;
@@ -17,7 +18,7 @@ export class ConversationManager {
       await fs.mkdir(this.conversationsDir, { recursive: true });
       await this.loadConversations();
     } catch (error) {
-      console.error('Failed to initialize conversation manager:', error);
+      await logOutput('error', 'Failed to initialize conversation manager', { error: (error as Error).message });
     }
   }
 
@@ -34,11 +35,11 @@ export class ConversationManager {
         }
       }
     } catch (error) {
-      console.error('Failed to load conversations:', error);
+      await logOutput('error', 'Failed to load conversations', { error: (error as Error).message });
     }
   }
 
-  async createConversation(title?: string): Promise<Conversation> {
+  async createConversation(_title?: string): Promise<Conversation> {
     const conversationId = `conv_${Date.now()}`;
     const now = Date.now();
 
@@ -122,7 +123,7 @@ export class ConversationManager {
     }
 
     this.currentConversationId = conversationId;
-    console.log(`✅ Switched to conversation: ${conversationId}`);
+    await logOutput('success', `�?Switched to conversation: ${conversationId}`);
   }
 
   listConversations(): Conversation[] {
@@ -146,7 +147,7 @@ export class ConversationManager {
       this.currentConversationId = null;
     }
 
-    console.log(`✅ Deleted conversation: ${conversationId}`);
+    await logOutput('success', `�?Deleted conversation: ${conversationId}`);
   }
 
   async clearCurrentConversation(): Promise<void> {
@@ -177,7 +178,7 @@ export class ConversationManager {
     const markdown = this.conversationToMarkdown(conversation);
     await fs.writeFile(outputPath, markdown, 'utf-8');
 
-    console.log(`✅ Exported conversation to: ${outputPath}`);
+    await logOutput('success', `�?Exported conversation to: ${outputPath}`);
   }
 
   private conversationToMarkdown(conversation: Conversation): string {
@@ -221,7 +222,7 @@ export class ConversationManager {
     this.conversations.set(conversation.id, conversation);
     await this.saveConversation(conversation);
 
-    console.log(`✅ Imported conversation: ${conversation.id}`);
+    await logOutput('success', `�?Imported conversation: ${conversation.id}`);
 
     return conversation;
   }
@@ -286,3 +287,4 @@ export function getConversationManager(): ConversationManager {
   }
   return conversationManagerInstance;
 }
+

package/src/gui-subagent/action-parser/actionParser.ts

@@ -302,7 +302,7 @@ function parseAction(actionStr: string) {
           value = `(${value})`;
         }
 
-        //@ts-ignore
+        //@ts-expect-error - kwargs type mismatch with function signature
         kwargs[key.trim()] = value;
       }
     }
@@ -311,7 +311,7 @@ function parseAction(actionStr: string) {
       function: functionName,
       args: kwargs,
     };
-  } catch (e) {
+  } catch {
     logger.debug(`[ActionParser] Skipping invalid action: '${actionStr}'`);
     return null;
   }

package/src/gui-subagent/agent/gui-agent.ts

@@ -9,20 +9,19 @@
 import type {
   ScreenContext,
   ScreenshotOutput,
-  ExecuteParams,
-  ExecuteOutput,
   PredictionParsed,
 } from '../types/operator.js';
 import type { Operator } from '../operator/base-operator.js';
 import { sleep, asyncRetry } from '../utils.js';
 import { actionParser } from '../action-parser/index.js';
-import { colors, icons, renderMarkdown } from '../../theme.js';
+import { colors, icons} from '../../theme.js';
 import { getLogger } from '../../logger.js';
+import { SdkOutputAdapter } from '../../sdk-output-adapter.js';
 
 /**
  * Helper function to truncate long text
  */
-function truncateText(text: string, maxLength: number = 200): string {
+function _truncateText(text: string, maxLength: number = 200): string {
   if (!text) return '';
   return text.length > maxLength ? text.substring(0, maxLength) + '...' : text;
 }
@@ -30,7 +29,7 @@ function truncateText(text: string, maxLength: number = 200): string {
 /**
  * Helper function to indent multiline text
  */
-function indentMultiline(text: string, indent: string): string {
+function _indentMultiline(text: string, indent: string): string {
   return text.split('\n').map(line => indent + line).join('\n');
 }
 
@@ -90,6 +89,11 @@ export interface GUIAgentConfig<T extends Operator> {
   maxLoopCount?: number;
   logger?: any;
   signal?: AbortSignal;
+  /**
+   * SDK output adapter for SDK mode output
+   * When provided, GUI Agent will use it to output status and progress in SDK format
+   */
+  sdkOutputAdapter?: SdkOutputAdapter | null;
   onData?: (data: GUIAgentData) => void;
   onError?: (error: Error) => void;
   showAIDebugInfo?: boolean;
@@ -156,6 +160,7 @@ export class GUIAgent<T extends Operator> {
   private readonly maxLoopCount: number;
   private readonly logger: Console;
   private readonly signal?: AbortSignal;
+  private readonly sdkOutputAdapter?: SdkOutputAdapter | null;
   private readonly onData?: (data: GUIAgentData) => void;
   private readonly onError?: (error: Error) => void;
   private readonly showAIDebugInfo: boolean;
@@ -181,6 +186,7 @@ export class GUIAgent<T extends Operator> {
     this.maxLoopCount = config.maxLoopCount || MAX_LOOP_COUNT;
     this.logger = config.logger || guiLogger;
     this.signal = config.signal;
+    this.sdkOutputAdapter = config.sdkOutputAdapter ?? null;
     this.onData = config.onData;
     this.onError = config.onError;
     this.showAIDebugInfo = config.showAIDebugInfo ?? false;
@@ -200,11 +206,14 @@ export class GUIAgent<T extends Operator> {
 
   /**
    * Display conversation results with formatting similar to session.ts (simplified)
+   * In SDK mode, uses the SDK adapter for structured output
+   * Note: For assistant actions, SDK output is handled in the action execution loop
+   * to ensure accurate timing information
    */
   private displayConversationResult(conversation: Conversation, iteration: number, indentLevel: number = 1): void {
     const indent = '  '.repeat(indentLevel);
     const innerIndent = '  '.repeat(indentLevel + 1);
-    const maxWidth = process.stdout.columns || 80;
+    const _maxWidth = process.stdout.columns || 80;
 
     if (conversation.from === 'assistant') {
       // Display assistant response (action)
@@ -215,16 +224,21 @@ export class GUIAgent<T extends Operator> {
       const actionSummary = content.replace(/Thought:[\s\S]*?Action:\s*/i, '').trim();
       const actionType = conversation.predictionParsed?.[0]?.action_type || 'action';
 
-      console.log(`${indent}${colors.primaryBright(`[${iteration}]`)} ${colors.textMuted(actionType)}${timing ? colors.textDim(` (${timing.cost}ms)`) : ''}`);
+      // In SDK mode, action output is handled in the action execution loop
+      // Only use console output for non-SDK mode
+      if (!this.sdkOutputAdapter) {
+        console.log(`${indent}${colors.primaryBright(`[${iteration}]`)} ${colors.textMuted(actionType)}${timing ? colors.textDim(` (${timing.cost}ms)`) : ''}`);
 
-      // Optionally show action details on next line if verbose
-      if (this.showAIDebugInfo && actionSummary) {
-        const truncatedSummary = actionSummary.length > 60 ? actionSummary.substring(0, 60) + '...' : actionSummary;
-        console.log(`${innerIndent}${colors.textMuted(truncatedSummary)}`);
+        // Optionally show action details on next line if verbose
+        if (this.showAIDebugInfo && actionSummary) {
+          const truncatedSummary = actionSummary.length > 60 ? actionSummary.substring(0, 60) + '...' : actionSummary;
+          console.log(`${innerIndent}${colors.textMuted(truncatedSummary)}`);
+        }
       }
     } else if (conversation.from === 'human' && conversation.screenshotBase64) {
       // Show minimal indicator for screenshot
-      if (this.showAIDebugInfo) {
+      // In SDK mode, screenshot is handled by the conversation data
+      if (this.showAIDebugInfo && !this.sdkOutputAdapter) {
         const timing = conversation.timing;
         console.log(`${indent}${colors.textMuted(`${icons.loading} screenshot${timing ? ` (${timing.cost}ms)` : ''}`)}`);
       }
@@ -240,18 +254,30 @@ export class GUIAgent<T extends Operator> {
 
     switch (status) {
       case GUIAgentStatus.RUNNING:
-        console.log(`${indent}${colors.info(`${icons.loading} Step ${iteration}: Running...`)}`);
+        if (!this.sdkOutputAdapter) {
+          console.log(`${indent}${colors.info(`${icons.loading} Step ${iteration}: Running...`)}`);
+        } else {
+          this.sdkOutputAdapter.outputInfo(`Step ${iteration}: Running...`);
+        }
         break;
       case GUIAgentStatus.END:
         // Handled by caller
         break;
       case GUIAgentStatus.ERROR:
         if (data.error) {
-          console.log(`${indent}${colors.error(`${icons.cross} ${data.error}`)}`);
+          if (!this.sdkOutputAdapter) {
+            console.log(`${indent}${colors.error(`${icons.cross} ${data.error}`)}`);
+          } else {
+            this.sdkOutputAdapter.outputError(data.error);
+          }
         }
         break;
       case GUIAgentStatus.USER_STOPPED:
-        console.log(`${indent}${colors.warning(`${icons.warning} Stopped`)}`);
+        if (!this.sdkOutputAdapter) {
+          console.log(`${indent}${colors.warning(`${icons.warning} Stopped`)}`);
+        } else {
+          this.sdkOutputAdapter.outputWarning('Stopped');
+        }
         break;
       default:
         break;
@@ -259,6 +285,7 @@ export class GUIAgent<T extends Operator> {
   }
 
   private buildSystemPrompt(): string {
+    /* eslint-disable no-useless-escape */
     return `You are a GUI agent. You are given a task and your action history, with screenshots. You need to perform the next action to complete the task.
 
 ## Output Format
@@ -273,7 +300,7 @@ left_double(point='<point>x1 y1</point>')
 right_single(point='<point>x1 y1</point>')
 drag(start_point='<point>x1 y1</point>', end_point='<point>x2 y2</point>')
 hotkey(key='ctrl c') # Split keys with a space and use lowercase. Also, do not use more than 3 keys in one hotkey action.
-type(content='xxx') # Use escape characters \', \", and \n in content part to ensure we can parse the content in normal python string format. If you want to submit your input, use \n at the end of content. 
+type(content='xxx') # Use escape characters \', \", and \n in content part to ensure we can parse the content in normal python string format. If you want to submit your input, use \n at the end of content.
 scroll(point='<point>x1 y1</point>', direction='down or up or right or left') # Show more information on the \`direction\` side.
 open_url(url='https://xxx') # Open URL in browser
 wait() #Sleep for 5s and take a screenshot to check for any changes.
@@ -287,6 +314,7 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
 - Write a small plan and finally summarize your next action (with its target element) in one sentence in \`Thought\` part.
 
 `;
+    /* eslint-enable no-useless-escape */
   }
 
 
@@ -315,6 +343,11 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
       ],
     };
 
+    // Output start via SDK adapter if available
+    if (this.sdkOutputAdapter) {
+      this.sdkOutputAdapter.outputGUIAgentStart(instruction, this.isLocalMode ? 'local' : 'remote');
+    }
+
     // Initialize operator for initial screenshot
     try {
       await this.operator.doInitialize();
@@ -333,10 +366,15 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
         data.status = GUIAgentStatus.ERROR;
         data.error = `Failed to initialize operator: ${errorMsg}`;
       }
+
+      // Output error via SDK adapter if available
+      if (this.sdkOutputAdapter) {
+        this.sdkOutputAdapter.outputGUIAgentError(data.error, errorMsg);
+      }
       return data;
     }
 
-    const currentTime = Date.now();
+    const _currentTime = Date.now();
 
     if (this.showAIDebugInfo) {
       this.logger.debug('[GUIAgent] run:', {
@@ -353,8 +391,20 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
     data.status = GUIAgentStatus.RUNNING;
     data.systemPrompt = this.systemPrompt;
     const indent = '  '.repeat(this.indentLevel);
-    console.log(`${indent}${colors.primaryBright(`${icons.rocket} GUI Agent started`)}`);
-    console.log('');
+
+    // Output start via SDK adapter if available, otherwise use console
+    if (this.sdkOutputAdapter) {
+      this.sdkOutputAdapter.outputGUIAgentStart(data.conversations[0]?.value || '', this.isLocalMode ? 'local' : 'remote');
+    } else {
+      console.log(`${indent}${colors.primaryBright(`${icons.rocket} GUI Agent started`)}`);
+      console.log('');
+    }
+
+    // Output running status via SDK adapter if available
+    if (this.sdkOutputAdapter) {
+      this.sdkOutputAdapter.outputGUIAgentStatus(GUIAgentStatus.RUNNING);
+    }
+
     await this.onData?.({ ...data, conversations: [] });
 
     try {
@@ -367,9 +417,17 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
         // Check pause status
         if (this.isPaused && this.resumePromise) {
           data.status = GUIAgentStatus.PAUSE;
+          // Output pause status via SDK adapter if available
+          if (this.sdkOutputAdapter) {
+            this.sdkOutputAdapter.outputGUIAgentStatus(GUIAgentStatus.PAUSE, loopCnt);
+          }
           await this.onData?.({ ...data, conversations: [] });
           await this.resumePromise;
           data.status = GUIAgentStatus.RUNNING;
+          // Output running status via SDK adapter if available
+          if (this.sdkOutputAdapter) {
+            this.sdkOutputAdapter.outputGUIAgentStatus(GUIAgentStatus.RUNNING, loopCnt);
+          }
           await this.onData?.({ ...data, conversations: [] });
         }
 
@@ -687,6 +745,12 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
 
                 // Any other status (success, failed, etc.) is considered success
                 stepSuccess = true;
+
+                // Output action via SDK adapter if available
+                if (this.sdkOutputAdapter && actionType) {
+                  const timingCost = Date.now() - start;
+                  this.sdkOutputAdapter.outputGUIAgentAction(loopCnt, actionType, timingCost);
+                }
                 break;
               } catch (executeError) {
                 stepRetryCount++;
@@ -767,7 +831,11 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
 
       // Output error immediately if task failed
       if (finalStatus === GUIAgentStatus.ERROR && finalError) {
-        console.log(`\n${indent}${colors.error('✖')} ${finalError}\n`);
+        if (!this.sdkOutputAdapter) {
+          console.log(`\n${indent}${colors.error('✖')} ${finalError}\n`);
+        } else {
+          this.sdkOutputAdapter.outputError(finalError);
+        }
       }
 
       // Call onData callback if set
@@ -796,6 +864,26 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
       // Log final status (only visible when showAIDebugInfo is enabled)
       this.logger.debug(`[GUIAgent] Final status: ${finalStatus}${finalError ? `, Error: ${finalError}` : ''}, Steps: ${loopCnt}`);
 
+      // Output final status via SDK adapter if available
+      if (this.sdkOutputAdapter) {
+        switch (finalStatus) {
+          case GUIAgentStatus.END:
+            this.sdkOutputAdapter.outputGUIAgentComplete(data.conversations[0]?.value || '', loopCnt);
+            break;
+          case GUIAgentStatus.USER_STOPPED:
+            this.sdkOutputAdapter.outputGUIAgentCancelled(data.conversations[0]?.value || '');
+            break;
+          case GUIAgentStatus.ERROR:
+            this.sdkOutputAdapter.outputGUIAgentError(
+              data.conversations[0]?.value || 'GUI Agent error',
+              finalError || 'Unknown error'
+            );
+            break;
+          default:
+            this.sdkOutputAdapter.outputGUIAgentStatus(finalStatus, loopCnt, finalError);
+        }
+      }
+
       data.status = finalStatus;
       data.error = finalError;
     }
@@ -1003,20 +1091,15 @@ finished(content='xxx') # Use escape characters \', \", and \n in content part t
       this.debugRequest(messages);
     }
 
-    let response;
-    try {
-      response = await fetch(`${baseUrl}/chat/completions`, {
-        method: 'POST',
-        headers: {
-          'Content-Type': 'application/json',
-          'Authorization': `Bearer ${apiKey}`,
-        },
-        body: JSON.stringify(requestBody),
-        signal: this.signal,
-      });
-    } catch (fetchError) {
-      throw fetchError;
-    }
+    const response = await fetch(`${baseUrl}/chat/completions`, {
+      method: 'POST',
+      headers: {
+        'Content-Type': 'application/json',
+        'Authorization': `Bearer ${apiKey}`,
+      },
+      body: JSON.stringify(requestBody),
+      signal: this.signal,
+    });
 
     // Handle non-200 responses
     if (!response.ok) {

package/src/gui-subagent/index.ts

@@ -21,6 +21,7 @@ import { GUIAgent, type GUIAgentConfig, type GUIAgentData, type Conversation, GU
 import type { Operator } from './operator/base-operator.js';
 import type { RemoteVlmCaller } from './agent/gui-agent.js';
 import { getCancellationManager } from '../cancellation.js';
+import { SdkOutputAdapter } from '../sdk-output-adapter.js';
 
 /**
  * GUI Subagent configuration
@@ -56,6 +57,11 @@ export interface GUISubAgentConfig {
   maxLoopCount?: number;
   showAIDebugInfo?: boolean;
   indentLevel?: number;
+  /**
+   * SDK output adapter for SDK mode output
+   * When provided, GUI Agent will use it to output status and progress in SDK format
+   */
+  sdkOutputAdapter?: SdkOutputAdapter | null;
 }
 
 /**
@@ -73,6 +79,7 @@ export const DEFAULT_GUI_CONFIG = {
   maxLoopCount: 100,
   showAIDebugInfo: false,
   indentLevel: 1,
+  sdkOutputAdapter: null,
 };
 
 /**
@@ -116,6 +123,7 @@ export async function createGUISubAgent<T extends Operator>(
     showAIDebugInfo: mergedConfig.showAIDebugInfo,
     indentLevel: mergedConfig.indentLevel,
     signal: abortController.signal,
+    sdkOutputAdapter: mergedConfig.sdkOutputAdapter ?? null,
   };
 
   const agent = new GUIAgent<T>(agentConfig);

package/src/gui-subagent/operator/computer-operator.ts

@@ -11,7 +11,6 @@ import {
   Button,
   Key,
   Point,
-  centerOf,
   keyboard,
   mouse,
   sleep,
@@ -197,6 +196,8 @@ export class ComputerOperator extends Operator {
     context: { startX: number; startY: number; screenWidth: number; screenHeight: number; scaleFactor: number }
   ): Promise<'end' | void> {
     const { startX, startY, screenWidth, screenHeight, scaleFactor } = context;
+    // eslint-disable-next-line @typescript-eslint/no-unused-vars
+    void scaleFactor;
 
     const moveStraightTo = async (x: number, y: number) => {
       await mouse.move(straightTo(new Point(x, y)));

package/src/input-processor.ts

@@ -74,7 +74,7 @@ export class InputProcessor {
     return imageInputs;
   }
 
-  private async getImageFromClipboard(imageId: string): Promise<string> {
+  private async getImageFromClipboard(_imageId: string): Promise<string> {
     try {
       const clipboardContent = await clipboardy.read();
       
@@ -84,7 +84,7 @@ export class InputProcessor {
 
       const imageData = await this.readImageFile(clipboardContent);
       return imageData;
-    } catch (error) {
+    } catch {
       throw new Error('Failed to read image from clipboard');
     }
   }

package/src/logger.ts

@@ -1,8 +1,6 @@
-import chalk from 'chalk';
 import wrapAnsi from 'wrap-ansi';
 import stringWidth from 'string-width';
-import stripAnsi from 'strip-ansi';
-import { theme, icons, colors, styleHelpers } from './theme.js';
+import { icons, colors, styleHelpers } from './theme.js';
 
 export enum LogLevel {
   ERROR = 'error',
@@ -232,7 +230,7 @@ export class Logger {
       return Math.max(stringWidth(header), maxRowWidth);
     });
 
-    const totalWidth = columnWidths.reduce((sum, width) => sum + width + 2, 0) + (columnWidths.length - 1);
+    const _totalWidth = columnWidths.reduce((sum, width) => sum + width + 2, 0) + (columnWidths.length - 1);
 
     const createSeparator = (left: string, middle: string, right: string, horizontal: string) => {
       return left + columnWidths.map(width => horizontal.repeat(width + 2)).join(middle) + right;