@xagent-ai/cli 1.2.2 → 1.3.1

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

@@ -1,106 +1,106 @@
-/**
- * Operator Types for gui-subagent
- * Based on UI-TARS architecture for cross-platform computer control
- */
-
-import type { SupportedActionType } from './actions.js';
-
-export type { SupportedActionType };
-
-/**
- * Status Enum from @ui-tars/sdk/core
- * Aligned with UI-TARS SDK StatusEnum
- */
-export enum StatusEnum {
-  SUCCESS = 'success',
-  FAILED = 'failed',
-  END = 'end',
-  CONTINUE = 'continue',
-}
-
-export interface ScreenContext {
-  /** Screenshot width */
-  width: number;
-  /** Screenshot height */
-  height: number;
-  /** Device DPR */
-  scaleFactor: number;
-}
-
-export interface ScreenshotOutput {
-  status: 'success' | 'failed';
-  base64?: string;
-  url?: string;
-  errorMessage?: string;
-  scaleFactor?: number;
-}
-
-export interface ExecuteParams {
-  prediction: string;
-  parsedPrediction: PredictionParsed;
-  /** Device Physical Resolution */
-  screenWidth: number;
-  /** Device Physical Resolution */
-  screenHeight: number;
-  /** Device DPR */
-  scaleFactor: number;
-  /** Model coordinates scaling factor [widthFactor, heightFactor] */
-  factors: [number, number];
-}
-
-export interface ExecuteOutput {
-  status: 'success' | 'failed' | 'end';
-  errorMessage?: string;
-}
-
-export interface PredictionParsed {
-  reflection: string | null;
-  thought: string;
-  action_type: string;
-  action_inputs: Record<string, any>;
-}
-
-export interface OperatorConfig {
-  headless?: boolean;
-  browserPath?: string;
-  viewport?: {
-    width: number;
-    height: number;
-  };
-  deviceScaleFactor?: number;
-}
-
-export interface OperatorInterface {
-  doInitialize(): Promise<void>;
-  doScreenshot(): Promise<ScreenshotOutput>;
-  doExecute(params: ExecuteParams): Promise<ExecuteOutput>;
-  getSupportedActions(): SupportedActionType[];
-  getScreenContext(): Promise<ScreenContext>;
-  cleanup(): Promise<void>;
-  destroyInstance(): Promise<void>;
-}
-
-/**
- * Computer Operator specific types
- */
-export interface ComputerOperatorConfig {
-  /** Screenshot quality (0-100) */
-  screenshotQuality?: number;
-  /** Maximum screenshot retry attempts */
-  maxScreenshotRetries?: number;
-  /** Delay between retries (ms) */
-  screenshotRetryDelay?: number;
-}
-
-/**
- * Operator Manual - describes supported actions and parameters
- */
-export interface OperatorManual {
-  ACTION_SPACES: string[];
-  KEY_SPACE?: {
-    [key: string]: string;
-  };
-  KEY_DESCRIPTION?: {
-    [key: string]: string;
-  };
-}
+/**
+ * Operator Types for gui-subagent
+ * Based on UI-TARS architecture for cross-platform computer control
+ */
+
+import type { SupportedActionType } from './actions.js';
+
+export type { SupportedActionType };
+
+/**
+ * Status Enum from @ui-tars/sdk/core
+ * Aligned with UI-TARS SDK StatusEnum
+ */
+export enum StatusEnum {
+  SUCCESS = 'success',
+  FAILED = 'failed',
+  END = 'end',
+  CONTINUE = 'continue',
+}
+
+export interface ScreenContext {
+  /** Screenshot width */
+  width: number;
+  /** Screenshot height */
+  height: number;
+  /** Device DPR */
+  scaleFactor: number;
+}
+
+export interface ScreenshotOutput {
+  status: 'success' | 'failed';
+  base64?: string;
+  url?: string;
+  errorMessage?: string;
+  scaleFactor?: number;
+}
+
+export interface ExecuteParams {
+  prediction: string;
+  parsedPrediction: PredictionParsed;
+  /** Device Physical Resolution */
+  screenWidth: number;
+  /** Device Physical Resolution */
+  screenHeight: number;
+  /** Device DPR */
+  scaleFactor: number;
+  /** Model coordinates scaling factor [widthFactor, heightFactor] */
+  factors: [number, number];
+}
+
+export interface ExecuteOutput {
+  status: 'success' | 'failed' | 'end' | 'needs_input';
+  errorMessage?: string;
+}
+
+export interface PredictionParsed {
+  reflection: string | null;
+  thought: string;
+  action_type: string;
+  action_inputs: Record<string, any>;
+}
+
+export interface OperatorConfig {
+  headless?: boolean;
+  browserPath?: string;
+  viewport?: {
+    width: number;
+    height: number;
+  };
+  deviceScaleFactor?: number;
+}
+
+export interface OperatorInterface {
+  doInitialize(): Promise<void>;
+  doScreenshot(): Promise<ScreenshotOutput>;
+  doExecute(params: ExecuteParams): Promise<ExecuteOutput>;
+  getSupportedActions(): SupportedActionType[];
+  getScreenContext(): Promise<ScreenContext>;
+  cleanup(): Promise<void>;
+  destroyInstance(): Promise<void>;
+}
+
+/**
+ * Computer Operator specific types
+ */
+export interface ComputerOperatorConfig {
+  /** Screenshot quality (0-100) */
+  screenshotQuality?: number;
+  /** Maximum screenshot retry attempts */
+  maxScreenshotRetries?: number;
+  /** Delay between retries (ms) */
+  screenshotRetryDelay?: number;
+}
+
+/**
+ * Operator Manual - describes supported actions and parameters
+ */
+export interface OperatorManual {
+  ACTION_SPACES: string[];
+  KEY_SPACE?: {
+    [key: string]: string;
+  };
+  KEY_DESCRIPTION?: {
+    [key: string]: string;
+  };
+}

package/src/gui-subagent/utils.ts

@@ -1,51 +1,51 @@
-/**
- * Utility functions for gui-subagent
- */
-
-/**
- * Sleep for a specified duration
- * @param ms - Duration in milliseconds
- */
-export function sleep(ms: number): Promise<void> {
-  return new Promise((resolve) => setTimeout(resolve, ms));
-}
-
-/**
- * Simple async retry utility with bail support
- * @param fn - Async function to retry (receives a bail function as first argument)
- * @param options - Retry options
- */
-export async function asyncRetry<T>(
-  fn: (bail: (error: Error) => void) => Promise<T>,
-  options: {
-    retries?: number;
-    minTimeout?: number;
-    onRetry?: (e: Error) => void;
-  } = {}
-): Promise<T> {
-  const { retries = 3, minTimeout = 1000, onRetry } = options;
-  let lastError: Error | undefined;
-
-  for (let attempt = 0; attempt <= retries; attempt++) {
-    try {
-      let bailCalled = false;
-      const bail = (error: Error) => {
-        bailCalled = true;
-        throw error;
-      };
-      const result = await fn(bail);
-      if (bailCalled) {
-        throw lastError;
-      }
-      return result;
-    } catch (error) {
-      lastError = error as Error;
-      if (attempt < retries) {
-        onRetry?.(lastError);
-        await sleep(minTimeout * Math.pow(2, attempt));
-      }
-    }
-  }
-
-  throw lastError;
-}
+/**
+ * Utility functions for gui-subagent
+ */
+
+/**
+ * Sleep for a specified duration
+ * @param ms - Duration in milliseconds
+ */
+export function sleep(ms: number): Promise<void> {
+  return new Promise((resolve) => setTimeout(resolve, ms));
+}
+
+/**
+ * Simple async retry utility with bail support
+ * @param fn - Async function to retry (receives a bail function as first argument)
+ * @param options - Retry options
+ */
+export async function asyncRetry<T>(
+  fn: (bail: (error: Error) => void) => Promise<T>,
+  options: {
+    retries?: number;
+    minTimeout?: number;
+    onRetry?: (e: Error) => void;
+  } = {}
+): Promise<T> {
+  const { retries = 3, minTimeout = 1000, onRetry } = options;
+  let lastError: Error | undefined;
+
+  for (let attempt = 0; attempt <= retries; attempt++) {
+    try {
+      let bailCalled = false;
+      const bail = (error: Error) => {
+        bailCalled = true;
+        throw error;
+      };
+      const result = await fn(bail);
+      if (bailCalled) {
+        throw lastError;
+      }
+      return result;
+    } catch (error) {
+      lastError = error as Error;
+      if (attempt < retries) {
+        onRetry?.(lastError);
+        await sleep(minTimeout * Math.pow(2, attempt));
+      }
+    }
+  }
+
+  throw lastError;
+}

package/src/index.ts

@@ -1,18 +1,17 @@
-export * from './types.js';
-export * from './config.js';
-export * from './auth.js';
-export * from './ai-client.js';
-export * from './tools.js';
-export * from './agents.js';
-export * from './memory.js';
-export * from './mcp.js';
-export * from './checkpoint.js';
-export * from './slash-commands.js';
-export * from './session.js';
-export * from './session-manager.js';
-export * from './workflow.js';
-export * from './input-processor.js';
-export * from './conversation.js';
-export * from './update.js';
-export * from './cancellation.js';
-export * from './context-compressor.js';
+export * from './types.js';
+export * from './config.js';
+export * from './auth.js';
+export { createAIClient, type AIClientInterface, createAIProvider, getAllModels, getProviderTypes } from './ai-client-factory.js';
+export * from './agents.js';
+export * from './memory.js';
+export * from './mcp.js';
+export * from './checkpoint.js';
+export * from './slash-commands.js';
+export * from './session.js';
+export * from './session-manager.js';
+export * from './workflow.js';
+export * from './input-processor.js';
+export * from './conversation.js';
+export * from './update.js';
+export * from './cancellation.js';
+export * from './context-compressor.js';

package/src/input-processor.ts

@@ -67,14 +67,14 @@ export class InputProcessor {
           }
         });
       } catch (error) {
-        console.warn(`Failed to process image #${imageId}:`, error);
+        console.warn(`[input-processor] Failed to process image #${imageId}: ${error instanceof Error ? error.message : String(error)}`);
       }
     }
 
     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');
     }
   }
@@ -101,7 +101,8 @@ export class InputProcessor {
       
       return `data:${mimeType};base64,${base64}`;
     } catch (error) {
-      throw new Error(`Failed to read image file: ${filePath}`);
+      const errorMsg = error instanceof Error ? error.message : String(error);
+      throw new Error(`Failed to read image file ${filePath}: ${errorMsg}`);
     }
   }
 
@@ -204,6 +205,7 @@ export class InputProcessor {
         content: clipboardContent
       }];
     } catch (error) {
+      console.warn(`[input-processor] Failed to read clipboard: ${error instanceof Error ? error.message : String(error)}`);
       return [];
     }
   }
@@ -251,7 +253,8 @@ export class InputProcessor {
       
       return `data:${mimeType};base64,${base64}`;
     } catch (error) {
-      throw new Error(`Failed to convert image to base64: ${filePath}`);
+      const errorMsg = error instanceof Error ? error.message : String(error);
+      throw new Error(`Failed to convert image to base64 ${filePath}: ${errorMsg}`);
     }
   }