Skip to content

AiTranslator

AiTranslator understands your natural language commands—ranging from single test steps to complete scenarios—and transforms them into executable sequences of SmartActions. Acts as the semantic bridge between natural language intents and the automation framework, ensuring your automation heals itself when things change.

Action Summary

Action Description
DoConfigure Set common execution parameters and limitations.
DoConfigureControlTypes Configure control types available for the test session.
DoLint Lint a generated script (RVL file with AI steps) for potential problems.
DoObjectLocator Generate an improved XPath locator for an object in the repository using AI-driven DOM inspection.
DoTestStep Execute a single step within a multi-step test case, providing the agent with full test context including all steps, current position, and previously generated code.

Action Detail

DoConfigure

Set common execution parameters and limitations.

AiTranslator.DoConfigure(system_prompt, timeout, n_last_images, max_tokens, token_limit, vendor, model)

Parameters:

Name Type Description
system_prompt string Additional prompt to be used for all interactions.
timeout number Maximum time to wait, in milliseconds
Optional, Default: "600000".
n_last_images number Number of images to remember.
Optional, Default: "5".
max_tokens number Maximum amount of tokens per response.
Optional, Default: "1000".
token_limit number Token limit to be used for whole prompt.
Optional, Default: "1000000".
vendor string LLM provider to use for AiTranslator: bedrock or openai.
Optional, Default: "bedrock".
model string Model name to use (e.g., anthropic.claude-sonnet-4-20250514-v1:0, gpt-4o). If not specified, uses the default model from AI Dashboard settings.
Optional.

DoConfigureControlTypes

Configure control types available for the test session. Control types define which UI control recognition libraries are active (e.g., DomLightningGrid, Dom, DomLightningListBox).

AiTranslator.DoConfigureControlTypes(controlTypes)

Parameters:

Name Type Description
controlTypes string Comma-separated list of control types to activate (e.g., "DomLightningGrid,DomLightningListBox").

Returns:

true after successfully configuring the control types.

DoLint

Lint a generated script (RVL file with AI steps) for potential problems. This is a static analysis entry point — it does not execute any UI interactions.

AiTranslator.DoLint(rvlPath)

Parameters:

Name Type Description
rvlPath string Path to the RVL file with AI steps to lint for potential problems.

Returns:

LintResult object with actions (parsed actions) and violations (array of problems found).

DoObjectLocator

Generate an improved XPath locator for an object in the repository using AI-driven DOM inspection. The LLM inspects the live page DOM around the target element and proposes a more robust, semantically meaningful XPath based on your instructions.

Prerequisites: The browser must be open and the page containing the target element must be loaded.

AiTranslator.DoObjectLocator(object_id, prompt, timeout, max_tokens, token_limit, maxIterations, ancestorDepth)

Parameters:

Name Type Description
object_id objectId Object ID from the repository (e.g., "Table_Last_Row").
prompt string Instruction for how to generate/improve the xpath (e.g., "Reference the last row of the table").
timeout number Maximum time for the operation in milliseconds.
Optional, Default: "120000".
max_tokens number Maximum tokens per LLM response.
Optional, Default: "4000".
token_limit number Total token budget for the operation.
Optional, Default: "100000".
maxIterations number Maximum number of tool-use loop iterations.
Optional, Default: "10".
ancestorDepth number Default number of ancestor levels to inspect.
Optional, Default: "4".

Returns:

SeSDoActionResult with the new xpath value, or false on failure.

DoTestStep

Execute a single step within a multi-step test case, providing the agent with full test context including all steps, current position, and previously generated code.

AiTranslator.DoTestStep(testCase, currentStep, generatedCode, translatorContext, timeout, n_last_images, max_tokens, token_limit)

Parameters:

Name Type Description
testCase object Array of step descriptions for the full test case.
currentStep number 1-based index of the current step to execute.
generatedCode string SmartAction code generated by prior steps (pass empty string for first step).
Optional, Default: "".
translatorContext ITranslatorContext
timeout number Maximum time to wait, in milliseconds
Optional, Default: "600000".
n_last_images number Number of images to remember.
Optional, Default: "5".
max_tokens number Maximum amount of tokens per response.
Optional, Default: "1000".
token_limit number Token limit to be used for whole prompt.
Optional, Default: "1000000".

Returns:

true if the step was completed successfully; otherwise, it returns false.