@minded-ai/mindedjs 1.0.0-ec2-beta-20 → 1.0.0-ec2-beta-170

package/dist/browserTask/executeBrowserTask.py

@@ -148,6 +148,119 @@ class ScreenshotCapture:
             # Don't raise - continue execution
 
 
+class LogsCapture:
+    """Logs capture class that saves browser-use history logs to S3"""
+    
+    def __init__(self, config: Optional[Dict[str, Any]] = None):
+        self.config = config or {}
+        
+        # Import boto3 only when logs capture is needed
+        try:
+            import boto3
+            from botocore.exceptions import ClientError
+            self.boto3 = boto3
+            self.ClientError = ClientError
+        except ImportError:
+            raise ImportError(
+                "boto3 is required for logs capture. "
+                "Please install it with: uv pip install boto3 "
+                "or run: npx minded setup-local-operator"
+            )
+        
+        # S3 configuration with defaults
+        self.s3_bucket = self.config.get('s3_bucket', os.getenv('SCREENSHOT_S3_BUCKET', 'global-development-agentsforce'))
+        self.s3_prefix = self.config.get('s3_prefix', os.getenv('SCREENSHOT_S3_PREFIX', 'browser-use-runs-logs/'))
+        
+        # AWS region configuration
+        self.aws_region = self.config.get('aws_region', os.getenv('AWS_REGION', 'us-east-1'))
+        
+        # Ensure prefix ends with slash
+        if not self.s3_prefix.endswith('/'):
+            self.s3_prefix += '/'
+        
+        # Get session ID from config or generate one
+        self.session_id = self.config.get('session_id', datetime.now().strftime("%Y%m%d_%H%M%S_%f")[:-3])
+        
+        # Initialize S3 client with region
+        self.s3_client = self.boto3.client('s3', region_name=self.aws_region)
+        
+        # Track accumulated logs
+        self.log_entries: List[Dict[str, Any]] = []
+        self.step_counter = 0
+        
+        logger.info(f"📝 Logs capture ENABLED - will save history logs periodically")
+        logger.info(f"   S3 destination: s3://{self.s3_bucket}/{self.s3_prefix}")
+        logger.info(f"   Session ID: {self.session_id}")
+        logger.info(f"   AWS Region: {self.aws_region}")
+    
+    async def capture_logs(self, agent: Any) -> None:
+        """Capture logs at step end and upload to S3"""
+        try:
+            step_number = self.step_counter
+            self.step_counter += 1
+            
+            # Get history object from agent
+            history = agent.history
+            
+            # Extract relevant information from history
+            log_entry = {
+                'step': step_number,
+                'timestamp': datetime.now().isoformat(),
+                'urls': history.urls() if hasattr(history, 'urls') else [],
+                'action_names': history.action_names() if hasattr(history, 'action_names') else [],
+            }
+            
+            # Extract model outputs if available
+            try:
+                model_outputs = history.model_outputs() if hasattr(history, 'model_outputs') else []
+                log_entry['model_outputs'] = [
+                    {
+                        'memory': h.current_state.memory if hasattr(h, 'current_state') else None,
+                        'next_goal': h.current_state.next_goal if hasattr(h, 'current_state') else None,
+                        'thinking': h.current_state.thinking if hasattr(h, 'current_state') else None,
+                    }
+                    for h in model_outputs
+                ]
+            except Exception as e:
+                logger.debug(f"Could not extract model outputs: {e}")
+                log_entry['model_outputs'] = []
+            
+            # Add to accumulated logs
+            self.log_entries.append(log_entry)
+                        
+            # Upload to S3
+            s3_key = f"{self.s3_prefix}{self.session_id}/operator.log"
+            
+            try:
+                # Convert logs to JSON string
+                logs_json = json.dumps(self.log_entries, indent=2, default=str)
+                logs_bytes = logs_json.encode('utf-8')
+                
+                # Upload to S3 with KMS encryption
+                self.s3_client.put_object(
+                    Bucket=self.s3_bucket,
+                    Key=s3_key,
+                    Body=logs_bytes,
+                    ContentType='application/json',
+                    ServerSideEncryption='aws:kms',
+                    Tagging='retention=30d'
+                )
+                
+                s3_url = f"s3://{self.s3_bucket}/{s3_key}"
+                logger.info(f"✅ Logs uploaded: {s3_url} (step {step_number}, size: {len(logs_bytes)} bytes, total entries: {len(self.log_entries)})")
+                
+            except self.ClientError as e:
+                logger.error(f"❌ Failed to upload logs for step #{step_number} to S3")
+                logger.error(f"   Error: {str(e)}")
+                logger.error(f"   Bucket: {self.s3_bucket}, Key: {s3_key}")
+                # Continue execution even if logs upload fails
+                
+        except Exception as e:
+            step_num = getattr(self, 'step_counter', 'unknown')
+            logger.error(f"❌ Error capturing logs #{step_num}: {str(e)}")
+            # Don't raise - continue execution
+
+
 def create_pydantic_model_from_schema(output_schema: Optional[List[Dict[str, Any]]]) -> Optional[type[BaseModel]]:
   if not output_schema:
     return None
@@ -172,7 +285,12 @@ def create_pydantic_model_from_schema(output_schema: Optional[List[Dict[str, Any
 
 
 async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Optional[str] = None, 
+<<<<<<< HEAD
+               otp_secret: Optional[str] = None, screenshot_config: Optional[Dict[str, Any]] = None,
+               logs_config: Optional[Dict[str, Any]] = None):
+=======
                otp_secret: Optional[str] = None, screenshot_config: Optional[Dict[str, Any]] = None, folder_path: str = None):
+>>>>>>> ec2-support
   llm = ChatOpenAI(
     model="gpt-4.1",
     api_key=os.getenv("OPENAI_API_KEY"),
@@ -191,6 +309,7 @@ async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Opt
   
   # Initialize screenshot capture if enabled
   screenshot_capture = None
+  logs_capture = None
   on_step_end_hook = None
   
   if screenshot_config and screenshot_config.get('enabled', False):
@@ -202,14 +321,31 @@ async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Opt
     # Add session_id to config
     screenshot_config['session_id'] = session_id
     screenshot_capture = ScreenshotCapture(screenshot_config)
+  else:
+    logger.info("📷 Screenshot capture is DISABLED for this browser task")
+  
+  # Initialize logs capture if enabled
+  if screenshot_config and screenshot_config.get('enabled', False):
+    logger.info("-" * 50)
+    logger.info("🎯 Initializing logs capture for browser task")
+    logger.info(f"   Session: {session_id}")
+    logger.info("-" * 50)
     
-    # Create hook function for step end only
+    # Add session_id to config
+    screenshot_config['session_id'] = session_id
+    logs_capture = LogsCapture(screenshot_config)
+  else:
+    logger.info("📝 Logs capture is DISABLED for this browser task")
+  
+  # Create composite hook function for step end
+  if screenshot_capture or logs_capture:
     async def on_step_end(agent: Any) -> None:
-      await screenshot_capture.capture_screenshot(agent)
+      if screenshot_capture:
+        await screenshot_capture.capture_screenshot(agent)
+      if logs_capture:
+        await logs_capture.capture_logs(agent)
     
     on_step_end_hook = on_step_end
-  else:
-    logger.info("📷 Screenshot capture is DISABLED for this browser task")
 
   output_schema = None
   if output_schema_json:
@@ -298,12 +434,20 @@ if __name__ == '__main__':
     output_schema = payload.get('outputSchema')
     otp_secret = payload.get('otpSecret')
     screenshot_config = payload.get('screenshotConfig')
+<<<<<<< HEAD
+    logs_config = payload.get('logsConfig')
+=======
     folder_path = payload.get('folderPath')
+>>>>>>> ec2-support
 
     if not session_id or not cdp_url or not task or not folder_path:
       raise SystemExit("Missing required fields in JSON payload: sessionId, cdpUrl, task, folderPath")
 
     output_schema_json = json.dumps(output_schema) if output_schema is not None else None
+<<<<<<< HEAD
+    asyncio.run(main(session_id, cdp_url, task, output_schema_json, otp_secret, screenshot_config, logs_config))
+=======
     asyncio.run(main(session_id, cdp_url, task, output_schema_json, otp_secret, screenshot_config, folder_path))
+>>>>>>> ec2-support
   else:
     raise SystemExit("Usage: uv run executeBrowserTask.py; send a JSON payload via stdin")

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@minded-ai/mindedjs",
-  "version": "1.0.0-ec2-beta-20",
+  "version": "1.0.0-ec2-beta-170",
   "description": "MindedJS is a TypeScript library for building agents.",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",

package/src/browserTask/executeBrowserTask.py

@@ -148,6 +148,119 @@ class ScreenshotCapture:
             # Don't raise - continue execution
 
 
+class LogsCapture:
+    """Logs capture class that saves browser-use history logs to S3"""
+    
+    def __init__(self, config: Optional[Dict[str, Any]] = None):
+        self.config = config or {}
+        
+        # Import boto3 only when logs capture is needed
+        try:
+            import boto3
+            from botocore.exceptions import ClientError
+            self.boto3 = boto3
+            self.ClientError = ClientError
+        except ImportError:
+            raise ImportError(
+                "boto3 is required for logs capture. "
+                "Please install it with: uv pip install boto3 "
+                "or run: npx minded setup-local-operator"
+            )
+        
+        # S3 configuration with defaults
+        self.s3_bucket = self.config.get('s3_bucket', os.getenv('SCREENSHOT_S3_BUCKET', 'global-development-agentsforce'))
+        self.s3_prefix = self.config.get('s3_prefix', os.getenv('SCREENSHOT_S3_PREFIX', 'browser-use-runs-logs/'))
+        
+        # AWS region configuration
+        self.aws_region = self.config.get('aws_region', os.getenv('AWS_REGION', 'us-east-1'))
+        
+        # Ensure prefix ends with slash
+        if not self.s3_prefix.endswith('/'):
+            self.s3_prefix += '/'
+        
+        # Get session ID from config or generate one
+        self.session_id = self.config.get('session_id', datetime.now().strftime("%Y%m%d_%H%M%S_%f")[:-3])
+        
+        # Initialize S3 client with region
+        self.s3_client = self.boto3.client('s3', region_name=self.aws_region)
+        
+        # Track accumulated logs
+        self.log_entries: List[Dict[str, Any]] = []
+        self.step_counter = 0
+        
+        logger.info(f"📝 Logs capture ENABLED - will save history logs periodically")
+        logger.info(f"   S3 destination: s3://{self.s3_bucket}/{self.s3_prefix}")
+        logger.info(f"   Session ID: {self.session_id}")
+        logger.info(f"   AWS Region: {self.aws_region}")
+    
+    async def capture_logs(self, agent: Any) -> None:
+        """Capture logs at step end and upload to S3"""
+        try:
+            step_number = self.step_counter
+            self.step_counter += 1
+            
+            # Get history object from agent
+            history = agent.history
+            
+            # Extract relevant information from history
+            log_entry = {
+                'step': step_number,
+                'timestamp': datetime.now().isoformat(),
+                'urls': history.urls() if hasattr(history, 'urls') else [],
+                'action_names': history.action_names() if hasattr(history, 'action_names') else [],
+            }
+            
+            # Extract model outputs if available
+            try:
+                model_outputs = history.model_outputs() if hasattr(history, 'model_outputs') else []
+                log_entry['model_outputs'] = [
+                    {
+                        'memory': h.current_state.memory if hasattr(h, 'current_state') else None,
+                        'next_goal': h.current_state.next_goal if hasattr(h, 'current_state') else None,
+                        'thinking': h.current_state.thinking if hasattr(h, 'current_state') else None,
+                    }
+                    for h in model_outputs
+                ]
+            except Exception as e:
+                logger.debug(f"Could not extract model outputs: {e}")
+                log_entry['model_outputs'] = []
+            
+            # Add to accumulated logs
+            self.log_entries.append(log_entry)
+                        
+            # Upload to S3
+            s3_key = f"{self.s3_prefix}{self.session_id}/operator.log"
+            
+            try:
+                # Convert logs to JSON string
+                logs_json = json.dumps(self.log_entries, indent=2, default=str)
+                logs_bytes = logs_json.encode('utf-8')
+                
+                # Upload to S3 with KMS encryption
+                self.s3_client.put_object(
+                    Bucket=self.s3_bucket,
+                    Key=s3_key,
+                    Body=logs_bytes,
+                    ContentType='application/json',
+                    ServerSideEncryption='aws:kms',
+                    Tagging='retention=30d'
+                )
+                
+                s3_url = f"s3://{self.s3_bucket}/{s3_key}"
+                logger.info(f"✅ Logs uploaded: {s3_url} (step {step_number}, size: {len(logs_bytes)} bytes, total entries: {len(self.log_entries)})")
+                
+            except self.ClientError as e:
+                logger.error(f"❌ Failed to upload logs for step #{step_number} to S3")
+                logger.error(f"   Error: {str(e)}")
+                logger.error(f"   Bucket: {self.s3_bucket}, Key: {s3_key}")
+                # Continue execution even if logs upload fails
+                
+        except Exception as e:
+            step_num = getattr(self, 'step_counter', 'unknown')
+            logger.error(f"❌ Error capturing logs #{step_num}: {str(e)}")
+            # Don't raise - continue execution
+
+
 def create_pydantic_model_from_schema(output_schema: Optional[List[Dict[str, Any]]]) -> Optional[type[BaseModel]]:
   if not output_schema:
     return None
@@ -172,7 +285,12 @@ def create_pydantic_model_from_schema(output_schema: Optional[List[Dict[str, Any
 
 
 async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Optional[str] = None, 
+<<<<<<< HEAD
+               otp_secret: Optional[str] = None, screenshot_config: Optional[Dict[str, Any]] = None,
+               logs_config: Optional[Dict[str, Any]] = None):
+=======
                otp_secret: Optional[str] = None, screenshot_config: Optional[Dict[str, Any]] = None, folder_path: str = None):
+>>>>>>> ec2-support
   llm = ChatOpenAI(
     model="gpt-4.1",
     api_key=os.getenv("OPENAI_API_KEY"),
@@ -191,6 +309,7 @@ async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Opt
   
   # Initialize screenshot capture if enabled
   screenshot_capture = None
+  logs_capture = None
   on_step_end_hook = None
   
   if screenshot_config and screenshot_config.get('enabled', False):
@@ -202,14 +321,31 @@ async def main(session_id: str, cdp_url: str, task: str, output_schema_json: Opt
     # Add session_id to config
     screenshot_config['session_id'] = session_id
     screenshot_capture = ScreenshotCapture(screenshot_config)
+  else:
+    logger.info("📷 Screenshot capture is DISABLED for this browser task")
+  
+  # Initialize logs capture if enabled
+  if screenshot_config and screenshot_config.get('enabled', False):
+    logger.info("-" * 50)
+    logger.info("🎯 Initializing logs capture for browser task")
+    logger.info(f"   Session: {session_id}")
+    logger.info("-" * 50)
     
-    # Create hook function for step end only
+    # Add session_id to config
+    screenshot_config['session_id'] = session_id
+    logs_capture = LogsCapture(screenshot_config)
+  else:
+    logger.info("📝 Logs capture is DISABLED for this browser task")
+  
+  # Create composite hook function for step end
+  if screenshot_capture or logs_capture:
     async def on_step_end(agent: Any) -> None:
-      await screenshot_capture.capture_screenshot(agent)
+      if screenshot_capture:
+        await screenshot_capture.capture_screenshot(agent)
+      if logs_capture:
+        await logs_capture.capture_logs(agent)
     
     on_step_end_hook = on_step_end
-  else:
-    logger.info("📷 Screenshot capture is DISABLED for this browser task")
 
   output_schema = None
   if output_schema_json:
@@ -298,12 +434,20 @@ if __name__ == '__main__':
     output_schema = payload.get('outputSchema')
     otp_secret = payload.get('otpSecret')
     screenshot_config = payload.get('screenshotConfig')
+<<<<<<< HEAD
+    logs_config = payload.get('logsConfig')
+=======
     folder_path = payload.get('folderPath')
+>>>>>>> ec2-support
 
     if not session_id or not cdp_url or not task or not folder_path:
       raise SystemExit("Missing required fields in JSON payload: sessionId, cdpUrl, task, folderPath")
 
     output_schema_json = json.dumps(output_schema) if output_schema is not None else None
+<<<<<<< HEAD
+    asyncio.run(main(session_id, cdp_url, task, output_schema_json, otp_secret, screenshot_config, logs_config))
+=======
     asyncio.run(main(session_id, cdp_url, task, output_schema_json, otp_secret, screenshot_config, folder_path))
+>>>>>>> ec2-support
   else:
     raise SystemExit("Usage: uv run executeBrowserTask.py; send a JSON payload via stdin")