windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/AddDataToFile.java-43-55 (1)

43-55: ⚠️ Potential issue | 🟠 Major

Null check occurs after potential NPE.

If getValue() returns null, calling .toString() on lines 43-44 will throw NullPointerException before the null check on line 50 is reached. Both fileContents and filePath values should be validated for null before conversion.

🐛 Proposed fix with proper null validation

     try {
-        String content = fileContents.getValue().toString();
-        String path = filePath.getValue().toString();
-        
-        logger.info("Adding data to file: " + path);
-        logger.info("Data to add: " + content);
-        
-        // Validate file path
-        if (path == null || path.trim().isEmpty()) {
+        Object contentValue = fileContents.getValue();
+        Object pathValue = filePath.getValue();
+
+        // Validate inputs before conversion
+        if (pathValue == null || pathValue.toString().trim().isEmpty()) {
             setErrorMessage("File path cannot be empty");
             ScreenshotUtils.captureAndUploadScreenshot(testStepResult, 
                     "add_data_to_file_failure_screenshot", logger);
             return Result.FAILED;
         }
+
+        String path = pathValue.toString();
+        String content = contentValue != null ? contentValue.toString() : "";
+
+        logger.info("Adding data to file: " + path);
+        logger.info("Data to add: " + content);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/AddDataToFile.java`
around lines 43 - 55, The code calls fileContents.getValue().toString() and
filePath.getValue().toString() before validating for null; update AddDataToFile
so you first validate fileContents and filePath and their getValue() results for
null/empty and only then call toString() on them. Specifically, check
fileContents == null or fileContents.getValue() == null (and similarly for
filePath) and on failure call setErrorMessage(...) and
ScreenshotUtils.captureAndUploadScreenshot(testStepResult,
"add_data_to_file_failure_screenshot", logger) and return Result.FAILED; only
after these checks convert values to String and proceed, keeping
logger.info(...) and other logic unchanged.

windows_advanced_actions/src/main/java/com/testsigma/addons/util/StringCompareUtil.java-10-54 (1)

10-54: ⚠️ Potential issue | 🟠 Major

Null pointer exception risk if either input string is null.

If string1 or string2 is null, calling .equals(), .equalsIgnoreCase(), .contains(), or .toLowerCase() will throw a NullPointerException. Add null checks at the start of the method.

🛡️ Proposed fix to add null validation

     public boolean performOperation(String string1, String string2, String operation){
+        if (string1 == null || string2 == null) {
+            this.errorMessage = "Input strings cannot be null. string1: " + string1 + ", string2: " + string2;
+            return false;
+        }
         boolean equalsCheck = false;
         boolean containsCheck = false;

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/util/StringCompareUtil.java`
around lines 10 - 54, performOperation risks NullPointerException when string1
or string2 is null; add early null validation in performOperation to check both
inputs before the switch, set this.errorMessage (e.g., "Input string1 is null"
or "Input string2 is null" or a combined message) and return false if either is
null so no .equals/.contains/.toLowerCase calls run on null; update handling for
all branches (equals, equals ignore-case, contains, contains ignore-case) to
rely on the validated non-null inputs and preserve setting
this.successMessage/this.errorMessage and the boolean result logic.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ExtractDataFromScreen.java-125-141 (1)

125-141: ⚠️ Potential issue | 🟠 Major

Potential NullPointerException when JSON fields are missing.

JsonNode.get() returns null when the field doesn't exist. Calling asBoolean() or asText() on a null reference will throw NullPointerException. Use path() instead, which returns a MissingNode that safely handles default values.

🐛 Proposed fix using path()

     private AIResponse parseAIResponse(String aiResponse) {
         try {
             ObjectMapper mapper = new ObjectMapper();
             JsonNode jsonNode = mapper.readTree(aiResponse);

             AIResponse response = new AIResponse();
-            response.isQueryRelated = jsonNode.get("isQueryRelated").asBoolean();
-            response.extractedText = jsonNode.get("extracted Text").asText("");
-            response.additionalData = jsonNode.get("additional data").asText("");
+            response.isQueryRelated = jsonNode.path("isQueryRelated").asBoolean(false);
+            response.extractedText = jsonNode.path("extracted Text").asText("");
+            response.additionalData = jsonNode.path("additional data").asText("");

             return response;
         } catch (Exception e) {
             logger.warn("Failed to parse AI response JSON: " + e.getMessage());
             logger.warn("Raw AI response: " + aiResponse);
             return null;
         }
     }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ExtractDataFromScreen.java`
around lines 125 - 141, The parseAIResponse method can NPE when fields are
absent because jsonNode.get(...) may return null; change jsonNode.get(...) calls
to jsonNode.path(...) in parseAIResponse so MissingNode handles defaults (e.g.,
response.isQueryRelated = jsonNode.path("isQueryRelated").asBoolean();
response.extractedText = jsonNode.path("extracted Text").asText("");
response.additionalData = jsonNode.path("additional data").asText("");), keep
the same AIResponse field assignments and existing exception handling/logging
with logger unchanged.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnImage.java-58-61 (1)

58-61: ⚠️ Potential issue | 🟠 Major

Don’t log the screenshot upload URL.

testStepResult.getScreenshotUrl() is typically a presigned upload URL. Writing it at info level exposes temporary upload credentials in the step logs.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnImage.java`
around lines 58 - 61, The code logs a presigned upload URL
(testStepResult.getScreenshotUrl()) which exposes temporary credentials; in
ClickOnImage (where logger.info is called around
testStepResult.getScreenshotUrl() and before ocr.uploadFile) remove or redact
those logger.info calls so the full URL is not written to logs (either delete
the two logger.info lines that reference getScreenshotUrl() or replace them with
a non-sensitive message like "Uploading screenshot to storage" without including
the URL); ensure no other logging in the ClickOnImage flow outputs
testStepResult.getScreenshotUrl().

windows_advanced_actions/src/main/java/com/testsigma/addons/util/KeyboardUtils.java-60-100 (1)

60-100: ⚠️ Potential issue | 🟠 Major

Handle unmappable characters and supplementary code points before calling Robot.

This method accepts characters via Character.isLetter() that cannot be typed on any keyboard layout. Characters like é, à, ñ, and CJK ideographs pass isLetter() but KeyEvent.getExtendedKeyCodeForChar() returns VK_UNDEFINED when they lack a key mapping, leading to undefined Robot behavior. Additionally, supplementary Unicode code points (U+10000 and above) are represented as surrogate pairs in Java's char type; passing them one character at a time causes each surrogate to fail validation and throw IllegalArgumentException mid-input, leaving Robot in an inconsistent state.

Add early validation to reject unmappable characters or accept only BMP code points. For full Unicode support, accept String input and handle surrogate pairs correctly using Character.toCodePoint() and Character.codePointAt().

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/util/KeyboardUtils.java`
around lines 60 - 100, typeCharacter(Robot robot, char character) currently
assumes every Character.isLetter/isDigit has a valid keycode and that
single-char input covers all Unicode; add early validation to reject unmappable
or surrogate chars before touching the Robot: in typeCharacter check if
Character.isSurrogate(character) (or Character.isHighSurrogate/isLowSurrogate)
and throw IllegalArgumentException for supplementary code points, and after
computing keyCode via KeyEvent.getExtendedKeyCodeForChar(...) ensure keyCode !=
KeyEvent.VK_UNDEFINED and throw before calling robot.keyPress; apply the same
keyCode != VK_UNDEFINED guard for letter, digit and SPECIAL_CHAR_MAP branches
(SPECIAL_CHAR_MAP entries should still be used when present), or alternatively
change the API to accept a String and process code points with
Character.codePointAt/Character.toChars and Character.toCodePoint for full
Unicode support.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/EraseDataInFile.java-39-49 (1)

39-49: ⚠️ Potential issue | 🟠 Major

Same null-safety issue – check getValue() before calling toString().

If filePath.getValue() returns null, line 39 throws an NPE before the validation on line 44 is reached.

🐛 Proposed fix

         try {
-            String path = filePath.getValue().toString();
-            
-            logger.info("Erasing data from file: " + path);
-            
-            // Validate file path
-            if (path == null || path.trim().isEmpty()) {
+            Object value = filePath.getValue();
+            if (value == null || value.toString().trim().isEmpty()) {
                 setErrorMessage("File path cannot be empty");
                 ScreenshotUtils.captureAndUploadScreenshot(testStepResult, 
                         "erase_data_in_file_failure_screenshot", logger);
                 return Result.FAILED;
             }
+            
+            String path = value.toString();
+            logger.info("Erasing data from file: " + path);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/EraseDataInFile.java`
around lines 39 - 49, The code calls filePath.getValue().toString() which can
NPE if getValue() is null; first retrieve the value into a local (e.g., valueObj
= filePath.getValue()), check valueObj for null or empty before calling
toString(), and only then assign path = valueObj.toString(); if null/empty call
setErrorMessage and ScreenshotUtils.captureAndUploadScreenshot(testStepResult,
"erase_data_in_file_failure_screenshot", logger) and return Result.FAILED to
preserve the existing error handling flow.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/AddDataToClipboard.java-37-47 (1)

37-47: ⚠️ Potential issue | 🟠 Major

Null check is ineffective – toString() will throw NPE before reaching it.

On line 37, dataToCopy.getValue().toString() is called. If getValue() returns null, this throws a NullPointerException before the null check on line 42 is reached.

🐛 Proposed fix to check for null before toString()

         try {
-            String data = dataToCopy.getValue().toString();
-            
-            logger.info("Adding data to clipboard: " + data);
-            
-            // Validate data
-            if (data == null) {
+            Object value = dataToCopy.getValue();
+            if (value == null) {
                 setErrorMessage("Data to copy cannot be null");
                 ScreenshotUtils.captureAndUploadScreenshot(testStepResult, 
                         "add_data_to_clipboard_failure_screenshot", logger);
                 return Result.FAILED;
             }
+            
+            String data = value.toString();
+            logger.info("Adding data to clipboard: " + data);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/AddDataToClipboard.java`
around lines 37 - 47, The null check is after calling
dataToCopy.getValue().toString(), which can NPE; fix by checking
dataToCopy.getValue() for null before invoking toString(): first retrieve Object
val = dataToCopy.getValue(), if val is null call setErrorMessage(...) and
ScreenshotUtils.captureAndUploadScreenshot(testStepResult,
"add_data_to_clipboard_failure_screenshot", logger) and return Result.FAILED;
otherwise set String data = val.toString() and proceed with logger.info("Adding
data to clipboard: " + data). Ensure you reference dataToCopy.getValue(), data,
setErrorMessage, ScreenshotUtils.captureAndUploadScreenshot, logger,
testStepResult and Result.FAILED in the updated flow.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnImage.java-249-429 (1)

249-429: 🛠️ Refactor suggestion | 🟠 Major

Remove large commented-out code block.

Lines 249-429 contain a large block of commented-out code (the previous OCR-based implementation). This should be removed to improve code clarity. If you need to reference the old implementation, it can be found in version control history.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnImage.java`
around lines 249 - 429, Remove the large commented-out legacy OCR implementation
inside the ClickOnImage class (the block that contains the old
urlToFileConverter(...) and cleanupFile(...) implementations and the commented
execute/closing braces); delete that entire commented section so only the active
class code (current execute and any real helper methods) remains, and then clean
up any now-unused imports or leftover comment markers to keep ClickOnImage and
its real methods consistent.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnImage.java-174-177 (1)

174-177: ⚠️ Potential issue | 🟠 Major

Format string placeholders without arguments.

The error messages contain format placeholders (%s) but no arguments are provided to String.format(). This will print literal %s in the error message.

Suggested fix

         } catch (IOException e) {
-            logger.info("Exception occurred while performing visual test at %s : %s" + ExceptionUtils.getStackTrace(e));
-            setErrorMessage("Unable to perform visual testing for : <b> %s</b>");
-            throw new RuntimeException("Error occurred while performing visual test at  ");
+            logger.info("Exception occurred while performing visual test: " + ExceptionUtils.getStackTrace(e));
+            setErrorMessage("Unable to perform visual testing");
+            throw new RuntimeException("Error occurred while performing visual test", e);

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnImage.java`
around lines 174 - 177, The log and messages in ClickOnImage (catch block around
the IOException) use "%s" placeholders but aren't passing arguments; update the
logger.info, setErrorMessage, and the RuntimeException to use proper
String.format(...) with the intended arguments (e.g., the image path or context
and ExceptionUtils.getStackTrace(e) or include the exception as a parameter), or
remove the "%s" placeholders and concatenate properly; ensure logger.info
includes the exception details and setErrorMessage contains the target
identifier, and throw a RuntimeException with a clear formatted message (use the
same identifiers used in the original lines).

windows_advanced_actions/src/main/java/com/testsigma/addons/util/OCRUtils.java-544-561 (1)

544-561: ⚠️ Potential issue | 🟠 Major

OkHttp Response is not closed—potential resource leak.

The Response object from client.newCall(request).execute() must be closed to release the underlying connection. Even when response.body().string() is called, the response should be closed in all code paths.

🐛 Proposed fix using try-with-resources

             logger.info("Making OCR API call");
-            Response response = client.newCall(request).execute();
-
-            if (response.isSuccessful() && response.body() != null) {
-                String responseBody = response.body().string();
-                logger.info("OCR Response body: " + responseBody);
-
-                OCRResponse ocrResponse = mapper.readValue(responseBody, OCRResponse.class);
-
-                if (ocrResponse.hasError()) {
-                    throw new RuntimeException("OCR API returned error: " + ocrResponse.getError());
-                }
-                if (!ocrResponse.hasText()) {
-                    throw new RuntimeException("No text found in the image");
+            try (Response response = client.newCall(request).execute()) {
+                if (response.isSuccessful() && response.body() != null) {
+                    String responseBody = response.body().string();
+                    logger.info("OCR API call successful");
+
+                    OCRResponse ocrResponse = mapper.readValue(responseBody, OCRResponse.class);
+
+                    if (ocrResponse.hasError()) {
+                        throw new RuntimeException("OCR API returned error: " + ocrResponse.getError());
+                    }
+                    if (!ocrResponse.hasText()) {
+                        throw new RuntimeException("No text found in the image");
+                    }
+                    return ocrResponse.getText();
+                } else {
+                    throw new RuntimeException("OCR API call failed with status: " + response.code());
                 }
-                return ocrResponse.getText();
-            } else {
-                throw new RuntimeException("OCR API call failed with status: " + (response.code()));
             }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/util/OCRUtils.java`
around lines 544 - 561, The Response returned by
client.newCall(request).execute() in OCRUtils (the block using Response response
= client.newCall(request).execute()) is never closed, causing a resource leak;
wrap the execute() call in a try-with-resources (try (Response response =
client.newCall(request).execute()) { ... }) so the Response (and its body) is
always closed on all code paths, keep the existing logic that reads
response.body().string(), handle null body as before, and rethrow the same
RuntimeExceptions inside the try block.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnPositionRelativeToText.java-222-247 (1)

222-247: ⚠️ Potential issue | 🟠 Major

OkHttp Response not closed - potential resource leak.

The Response object from OkHttp must be closed to release the underlying connection. Currently, if the response is successful, the connection may not be properly released.

🐛 Proposed fix using try-with-resources

             logger.info("Making OCR API call");
-            Response response = client.newCall(request).execute();
-
-            if (response.isSuccessful()) {
-                logger.info("OCR API response is successful");
-                if (response.body() != null) {
-                    String responseBody = response.body().string();
+            try (Response response = client.newCall(request).execute()) {
+                if (response.isSuccessful()) {
+                    logger.info("OCR API response is successful");
+                    if (response.body() != null) {
+                        String responseBody = response.body().string();
                     // ... rest of processing
+                    } else {
+                        throw new RuntimeException("OCR API returned null response body");
+                    }
+                } else {
+                    throw new RuntimeException("OCR API call failed with status: " + response.code());
+                }
+            }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnPositionRelativeToText.java`
around lines 222 - 247, The Response from OkHttp (variable Response response
returned by client.newCall(request).execute() in ClickOnPositionRelativeToText)
is not closed and can leak connections; wrap the execute call in a
try-with-resources (e.g., try (Response response =
client.newCall(request).execute()) { ... }) or explicitly close the
response/body after reading, then read response.body().string() inside that
scope, deserialize with mapper into ocrResponse, and keep existing error/status
handling unchanged so the response is always closed even on exceptions.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnPositionRelativeToText.java-52-53 (1)

52-53: 🛠️ Refactor suggestion | 🟠 Major

Instance fields should be local variables.

ObjectMapper and OCRResponse are created as instance fields but used only within execute() and extractTextPoints(). The ocrResponse field is particularly problematic as it gets overwritten on each OCR call, which could cause issues if this action is reused.

♻️ Suggested refactor

-    ObjectMapper mapper = new ObjectMapper();
-    OCRResponse ocrResponse = new OCRResponse();
+    private static final ObjectMapper mapper = new ObjectMapper();

And make ocrResponse a local variable inside extractTextPoints().

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnPositionRelativeToText.java`
around lines 52 - 53, ObjectMapper and OCRResponse are declared as instance
fields but only used inside execute() and extractTextPoints(), and ocrResponse
is overwritten on each call; move their declarations into the methods that use
them: create a local ObjectMapper inside execute() (or inside
extractTextPoints() if only used there) and make ocrResponse a local variable
inside extractTextPoints(), updating references in
ClickOnPositionRelativeToText.execute() and
ClickOnPositionRelativeToText.extractTextPoints() accordingly so no shared
mutable state remains.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnTextWithWait.java-67-71 (1)

67-71: ⚠️ Potential issue | 🟠 Major

Handle the same null OCR case before logging it.

getTextPointFromText() already treats null as a normal "not found yet" case, but printAllCoordinates(textPoints) runs first and will throw on that same input. That turns a normal poll miss into an immediate failure.

Also applies to: 137-153

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnTextWithWait.java`
around lines 67 - 71, printAllCoordinates(textPoints) can NPE when OCR returns
null, turning a normal "not found yet" poll into a hard failure; change the
sequence in ClickOnTextWithWait so you call getTextPointFromText(textPoints) (or
check textPoints for null/empty) before calling printAllCoordinates, and only
invoke printAllCoordinates when textPoints is non-null and non-empty; apply the
same defensive null/empty check-update to the other similar block that uses
extractTextFromImage/getTextPointFromText/printAllCoordinates so both polling
paths handle a null OCR result gracefully.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java-70-74 (1)

70-74: ⚠️ Potential issue | 🟠 Major

Every poll leaks a temporary screenshot.

screenshotFile is created on each iteration, but there is no cleanup on miss, success, or exceptions. Longer waits will keep leaving PNGs behind in the temp directory.

Also applies to: 78-105, 123-135

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java`
around lines 70 - 74, The loop in ClickOnTextWithSpacesAndWait creates a
temporary screenshot via saveScreenshotToFile on every poll but never deletes
it, leaking PNGs; update the polling logic (the method in class
ClickOnTextWithSpacesAndWait that calls saveScreenshotToFile and
extractTextPoints) to ensure each screenshotFile is deleted after use in all
paths (miss, success, and exception) by moving deletion into a finally block or
equivalent try-with-resources cleanup; apply the same cleanup to the other
screenshot usages mentioned around the extractTextPoints call (the blocks
covering lines ~78-105 and ~123-135) so every temporary file is removed
regardless of outcome.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/WaitUntilTextPresentInScreen.java-81-88 (1)

81-88: ⚠️ Potential issue | 🟠 Major

Don't stop polling before the declared timeout.

When remainingTime drops below POLLING_INTERVAL_MS, this breaks immediately instead of waiting out the remainder and doing one last OCR check. That makes the action fail up to 1.5 seconds early for texts that appear near the deadline.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/WaitUntilTextPresentInScreen.java`
around lines 81 - 88, The loop exits early when remainingTime <
POLLING_INTERVAL_MS, causing a missed final OCR attempt; in
WaitUntilTextPresentInScreen replace the else { break } behavior by sleeping the
remainingTime (if >0) and then performing one last check instead of breaking —
use remainingTime = endTime - System.currentTimeMillis(); if (remainingTime > 0)
Thread.sleep(remainingTime) (handle InterruptedException) and then proceed to
run the final OCR check (or let the loop iterate once more) rather than breaking
out; reference POLLING_INTERVAL_MS, remainingTime, endTime, Thread.sleep and the
WaitUntilTextPresentInScreen polling loop.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnTextWithWait.java-90-98 (1)

90-98: ⚠️ Potential issue | 🟠 Major

The last sub-1.5s window is skipped.

As soon as remainingTime <= POLLING_INTERVAL_MS, this breaks instead of waiting out the remainder and retrying once more. That shortens the requested wait and makes the action flaky near the deadline.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnTextWithWait.java`
around lines 90 - 98, The loop in ClickOnTextWithWait uses endTime and
POLLING_INTERVAL_MS and breaks when remainingTime <= POLLING_INTERVAL_MS, which
skips the final partial wait; instead, compute remainingTime = endTime -
System.currentTimeMillis() and if remainingTime > 0 sleep for
Math.min(POLLING_INTERVAL_MS, remainingTime) (or sleep remainingTime if smaller)
and then retry once more; only break when remainingTime <= 0 so the full
requested timeout is honored.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/WaitUntilImagePresentWithFindImage.java-65-76 (1)

65-76: ⚠️ Potential issue | 🟠 Major

The polling loop never cleans up its temporary files.

Each iteration writes a new screenshot PNG, and urlToFileConverter() may also create a temp copy for remote image-url inputs. None of those files are deleted on success, timeout, or exception, so repeated waits will keep growing the temp directory. Be careful to only clean up files created by this action, not caller-supplied local paths.

Also applies to: 79-90, 173-207

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/WaitUntilImagePresentWithFindImage.java`
around lines 65 - 76, The polling loop creates temporary screenshot files and
urlToFileConverter("target_image", imageUrlValue) may create a temp copy for
remote inputs but none are deleted; update WaitUntilImagePresentWithFindImage to
track which files it creates (e.g., the per-iteration screenshotFile and a
boolean/marker for the searchImageFile returned by urlToFileConverter) and
delete those files on success, timeout, and on any exception (use try/finally
around the loop and ensure inside-loop cleanup of each screenshotFile after
processing each iteration). Ensure you only delete files that this action
created (do not delete caller-supplied local paths) by using the marker or by
detecting remote-generated temp files from urlToFileConverter before removing
them. Also apply the same cleanup pattern to the other affected blocks
referenced (lines ~79-90 and ~173-207) where temp files are created.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/VerifyTextInApplication.java-17-31 (1)

17-31: ⚠️ Potential issue | 🟠 Major

The result runtime variable is declared but never set.

The action text says it stores the outcome in result-variable-name, but testData1 is never assigned in either branch. Later steps cannot consume the verification result this action advertises.

Also applies to: 59-66

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/VerifyTextInApplication.java`
around lines 17 - 31, The runtime variable field testData1 (in class
VerifyTextInApplication) is never assigned; update the action's execution flow
(inside the execute method or the branches that determine text presence) to set
the runtime variable represented by testData1 to the boolean result (e.g.,
"true"/"false") after the OCR check. Use the TestData runtime API on testData1
(e.g., call the TestData runtime setter provided by the Testsigma SDK such as
setValue/setRuntimeValue or the project’s runtime variable helper) to persist
the outcome so later steps can read the variable.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/VerifyTextInApplication.java-24-37 (1)

24-37: ⚠️ Potential issue | 🟠 Major

The advertised runtime variable is never populated.

The action text says the result is stored in test-data1, but testData1 is never written before either return path. Downstream steps will always see that runtime variable unset.

Also applies to: 93-101

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/VerifyTextInApplication.java`
around lines 24 - 37, The runtime variable test-data1 is never assigned before
the method returns; update the VerifyTextInApplication execution flow (e.g., in
the execute() or equivalent method) to set testData1 with the result before
every return path — for example call testData1.setValue(String.valueOf(found))
or testData1.setValue(foundText) depending on expected value and then proceed
with the existing result handling; apply the same fix to the similar block
mentioned around lines 93-101 so all exit paths write to testData1.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java-107-116 (1)

107-116: ⚠️ Potential issue | 🟠 Major

This wait can exit up to one poll early.

When remainingTime <= POLLING_INTERVAL_MS, the loop breaks instead of waiting out the remainder and retrying once more. That shortens the advertised timeout and makes late-appearing text flaky.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java`
around lines 107 - 116, The loop can exit up to one poll early because it breaks
when remainingTime <= POLLING_INTERVAL_MS; instead, compute a sleep duration as
the lesser of POLLING_INTERVAL_MS and remainingTime and sleep that amount if
remainingTime > 0, only breaking when remainingTime <= 0. Update the block in
ClickOnTextWithSpacesAndWait to use remainingTime/endTime and
POLLING_INTERVAL_MS to derive nextSleep = Math.min(POLLING_INTERVAL_MS,
remainingTime), log the actual sleep seconds, call Thread.sleep(nextSleep) when
remainingTime > 0, and only break when remainingTime <= 0 so the full timeout is
honored.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java-173-179 (1)

173-179: ⚠️ Potential issue | 🟠 Major

Treat "no text in this frame" as a miss, not a failure.

In a wait action, an OCR response with no text should just mean "keep polling." Throwing here aborts the step on the first blank frame instead of waiting until timeout.

🛠️ Suggested change

-                    if (!ocrResponse.hasText()) {
-                        throw new RuntimeException("No text found in the image");
-                    }
+                    if (!ocrResponse.hasText()) {
+                        logger.info("No OCR text detected in this poll attempt");
+                        return java.util.Collections.emptyList();
+                    }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithSpacesAndWait.java`
around lines 173 - 179, The current logic in ClickOnTextWithSpacesAndWait
erroneously throws on ocrResponse.hasText() == false, causing the wait action to
abort on the first blank frame; change the behavior so only
ocrResponse.hasError() throws, while a missing text result is treated as a
non-fatal miss (continue polling until timeout) by returning to the polling loop
(or skipping this iteration) instead of throwing—update the code paths in
ClickOnTextWithSpacesAndWait where ocrResponse is checked so that
!ocrResponse.hasText() does not raise an exception but allows the wait loop to
keep retrying.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java-73-73 (1)

73-73: ⚠️ Potential issue | 🟠 Major

Avoid logging raw OCR text from the entire screen.

printAllCoordinates() emits every recognized string and coordinate at info, which can leak unrelated UI content into step logs. Log the match target or element count instead.

Also applies to: 217-220

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java`
at line 73, The call to printAllCoordinates(textPoints) in
ClickOnPositionRelativeToText logs all OCR-recognized strings and coordinates
and may leak unrelated UI content; replace these info-level dumps with minimal
safe logs that record only the search target and counts (e.g., "Found N matches
for '<targetText>'") and/or the chosen coordinate used for the click. Update
both occurrences that call printAllCoordinates (the one shown and the other call
around the 217-220 region) to stop printing raw OCR text and instead log the
target string and textPoints.size() and the final selected Point.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/VerifyTextInApplicationWithAI.java-45-53 (1)

45-53: ⚠️ Potential issue | 🟠 Major

Clean up the temp screenshot after upload.

saveScreenshotToFile() creates a persistent temp PNG, but neither the success path nor the catch path removes it. Repeated steps will leave stale screenshots on the agent.

Also applies to: 65-69, 73-83

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/VerifyTextInApplicationWithAI.java`
around lines 45 - 53, The temp screenshot file created by saveScreenshotToFile
in VerifyTextInApplicationWithAI is never deleted; after calling
ScreenshotUtils.uploadScreenshotToS3 (and in any catch/failure branches) ensure
the File returned by saveScreenshotToFile is deleted (or deletedOnExit) to avoid
accumulating files. Update the method containing the snippet to wrap screenshot
creation/processing in a try/finally (or ensure cleanup in both success and
exception handlers) and call screenshotFile.delete() (or Files.deleteIfExists)
after upload and after any early returns so that saveScreenshotToFile,
OCRUtils.extractTextPoints, OCRUtils.searchForText, and
ScreenshotUtils.uploadScreenshotToS3 all run but the temporary file is always
removed afterward.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java-157-158 (1)

157-158: ⚠️ Potential issue | 🟠 Major

Remove raw OCR text from logs.

Logging the full OCR response body and all detected strings dumps full-screen text into step logs, which is a privacy/compliance risk. Log counts or the matched string only.

Also applies to: 218-221

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java`
around lines 157 - 158, The code currently logs the full OCR response body and
all detected strings (see logger.info("OCR Response body: " + responseBody) in
ClickOnTextWithWait and the similar logging at the later block around lines
218-221); remove these raw text dumps and replace them with privacy-safe logs
that only record summary info such as the number of detected text blocks and, if
needed, the single matched string (not the entire response). Update the logging
calls that use responseBody and any collections of detected strings to log
counts (e.g., detectedList.size()) and/or the matchedText variable only, and
ensure no full OCR payload or full list of detected strings is written to logs.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java-50-118 (1)

50-118: ⚠️ Potential issue | 🟠 Major

Honor wait-time-in-seconds in the execution path.

maxWaitSeconds is only used in log/error text here. The action does a single OCR pass and then fails immediately, so any text that appears after the first capture is missed.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java`
around lines 50 - 118, The current implementation in
ClickOnPositionRelativeToText does a single OCR pass and ignores maxWaitSeconds;
change the flow to poll until the timeout by wrapping the
screen-capture→saveScreenshotToFile→ocr.extractTextFromImage→getTextPointFromText
sequence in a loop that checks elapsed time against maxWaitSeconds (use
System.currentTimeMillis() or similar), sleeping a short interval between
attempts, and breaking when OCR returns a non-null OCRTextPoint; ensure
calculateClickPosition, robot actions and the temporary screenshotFile cleanup
happen only after a successful find, and always delete any temporary files
created in each iteration to avoid leaks; keep existing log messages but move
the timeout log/error and Result.FAILED return to after the polling loop so the
action honors wait-time-in-seconds.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java-71-72 (1)

71-72: ⚠️ Potential issue | 🟠 Major

Delete the temp screenshot on every polling iteration.

Each pass through the wait loop creates a new temp PNG and none of them are removed on retry, success, or exception. A long timeout will leak many files on the agent.

Also applies to: 97-119

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java`
around lines 71 - 72, The per-iteration temp PNG created via
saveScreenshotToFile(screenCapture, "click_text_screenshot") is never removed;
modify the polling loop in ClickOnTextWithWait so each iteration deletes the
temporary screenshot after use (including on success and on exceptions). Wrap
the per-iteration screenshot creation/processing in a try/finally (or ensure a
Files.deleteIfExists/screenshotFile.delete() is called) so the File returned
from saveScreenshotToFile is removed after processing; apply the same cleanup
for the other occurrences in the wait loop (the block covering the later
saveScreenshotToFile calls around lines 97-119).

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java-67-67 (1)

67-67: ⚠️ Potential issue | 🟠 Major

Delete the temp screenshot from a finally block.

The current delete() calls only run on the normal success/not-found branches. Any OCR or click failure after saveScreenshotToFile(...) leaks the PNG on the local agent.

Also applies to: 98-109

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsLite/ClickOnPositionRelativeToText.java`
at line 67, Declare and assign the File returned by saveScreenshotToFile(...)
(the screenshotFile variable) outside/above the try block that performs OCR and
clicking, and then move the deletion logic into a finally block so the temporary
PNG is always removed even if OCR or click throws; specifically, ensure in the
finally you check screenshotFile != null && screenshotFile.exists() and attempt
screenshotFile.delete() (with optional logging on failure), and apply the same
change to the other similar block referenced around lines 98-109 so both
saveScreenshotToFile usages always clean up their temp files.

windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java-163-169 (1)

163-169: ⚠️ Potential issue | 🟠 Major

Return an empty OCR result when nothing is detected.

Throwing here makes a blank/loading screen fail the step immediately, even though this action is supposed to keep polling until the wait time expires.

♻️ Suggested change

-                    if (!ocrResponse.hasText()) {
-                        throw new RuntimeException("No text found in the image");
-                    }
-
-                    return ocrResponse.getText();
+                    if (!ocrResponse.hasText()) {
+                        return java.util.Collections.emptyList();
+                    }
+
+                    return ocrResponse.getText();

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In
`@windows_advanced_actions/src/main/java/com/testsigma/addons/windowsAdvanced/ClickOnTextWithWait.java`
around lines 163 - 169, The code currently throws a RuntimeException when
ocrResponse.hasText() is false, causing the step to fail immediately; modify
ClickOnTextWithWait so that instead of throwing for missing text it returns an
empty/blank OCR result (an OCRResponse instance with no text or a clearly empty
marker) so the polling loop can continue until the wait expires; keep the
existing throw for ocrResponse.hasError() as-is, and replace the `throw new
RuntimeException("No text found in the image")` branch with a
return/empty-response path that the surrounding polling logic in
ClickOnTextWithWait can recognize and retry.