config.properties.example

# Basic agent configuration

# Set to 'true' to run the agent without user interaction.
unattended.mode=false

# The port on which the agent server will listen.
port=8005

# The host address the agent server will bind to.
host=localhost

# Set to 'true' to enable debug mode for testing purposes.
debug.mode=true

 #-----------------------------------------------
 # RAG config

 # The provider for the vector database. Currently, only 'chroma' is supported.
 vector.db.provider=chroma

 # The URL of the vector database.
 vector.db.url=http://localhost:8020

 # The number of top relevant results to retrieve from the vector database.
 retriever.top.n=5

 #-----------------------------------------------
 # Model Configuration

 # The AI model provider to use (e.g., google, openai, groq).
 model.provider=google

 # The name of the instruction model to use.
 instruction.model.name=gemini-2.5-flash

 # The name of the vision model to use.
 vision.model.name=gemini-2.5-flash

 # The maximum number of output tokens for the model.
 model.max.output.tokens=5000

 # The temperature for model generation (controls randomness).
 model.temperature=0.0

 # The top_p value for model generation (controls diversity).
 model.top.p=1.0

 # Max API request retries
 model.max.retries=10

 # Enable/disable model logging
 model.logging.enabled=false

 # Enable/disable thinking process output
 thinking.output.enabled=true

 # Budget for Gemini thinking process
 gemini.thinking.budget=0

 #-----------------------------------------------
 # Google API Configuration

 # The provider for Google API (e.g., studio_ai, vertex_ai).
 google.api.provider=studio_ai

 # Your Google AI token.
 google.api.token=YOUR_GOOGLE_AI_TOKEN_HERE

 # Your Google Cloud project ID.
 google.project=YOUR_GOOGLE_PROJECT_ID

 # Your Google Cloud project location.
 google.location=YOUR_GOOGLE_PROJECT_LOCATION

 #-----------------------------------------------
 # Azure OpenAI API Configuration

 # Your Azure OpenAI API key.
 azure.openai.api.key=YOUR_AZURE_OPEN_AI_API_KEY_HERE

 # Your Azure OpenAI endpoint.
 azure.openai.endpoint=YOUR_AZURE_OPEN_AI_ENDPOINT_HERE

 #-----------------------------------------------
 # Groq API Configuration

 # Your Groq API key.
 groq.api.key=YOUR_GROQ_API_KEY_HERE

 # Your Groq API endpoint.
 groq.endpoint=YOUR_GROQ_ENDPOINT_HERE

 #-----------------------------------------------
 # Timeout and Retry Configuration

 # Timeout in milliseconds for test step execution retries.
 test.step.execution.retry.timeout.millis=5000

 # Interval in milliseconds between test step execution retries.
 test.step.execution.retry.interval.millis=1000

 # Timeout in milliseconds for verification retries.
 verification.retry.timeout.millis=5000

 # Delay in milliseconds before performing action verification.
 action.verification.delay.millis=500

 #-----------------------------------------------
 # user UI dialog settings

 # Default horizontal gap for UI dialogs.
 dialog.default.horizontal.gap=10

 # Default vertical gap for UI dialogs.
 dialog.default.vertical.gap=10

 # Default font type for UI dialogs.
 dialog.default.font.type=Dialog

 # Interval in milliseconds to check for user interaction in dialogs.
 dialog.user.interaction.check.interval.millis=50

 # Default font size for UI dialogs.
 dialog.default.font.size=13

 #-----------------------------------------------
 # Element Configuration

 # The color used for drawing bounding boxes around UI elements.
 element.bounding.box.color=green

 # this score is the minimum score which needs to be reached during target UI element retrieval based on its name
 element.retrieval.min.target.score=0.8

 # this score is the minimum score which needs to be reached during retrieval of all UI elements which are at least partially similar to
 # the target element name - they could be later modified by user in order to enhance the retrieval results next time
 element.retrieval.min.general.score=0.5

 # this score is the minimum score which needs to be reached during retrieval of all UI elements which are at least partially similar to
 # the target element name - they could be later modified by user in order to enhance the retrieval results next time
 element.retrieval.min.page.relevance.score=0.5

 # this score defines the threshold value of pattern match identification by OpenCV library, top visual matches which reach this score
 # will be passed to the model in order get the best one which matches the visual appearance and textual description
 element.locator.visual.similarity.threshold=0.8

 # the max amount of matches by OpenCV library which will be selected for each target UI element screenshot. Those matches will be
 # passed to the model in order get the best one which matches the visual appearance and textual description
 element.locator.top.visual.matches=6

 # The minimum intersection area ratio for a visual match to be considered valid.
 # This parameter helps filter out false positives where the detected bounding box
 # only partially overlaps with the expected element. A value of 0.8 means that
 # at least 80% of the detected bounding box must overlap with the expected element's
 # area to be considered a valid match.
 element.locator.min.intersection.area.ratio=0.8

 # The maximum allowed deviation ratio for the dimensions (width and height) of a found visual match compared to the original element's dimensions.
 # This parameter is used in the ORB (Oriented FAST and Rotated BRIEF) feature detection algorithm to filter out potential matches that are significantly
 # different in size from the element being searched for. A value of 0.3 means that a found match can be at most 30% larger than the original element
 # in either width or height to be considered a valid match. This helps in improving the accuracy of visual searches by eliminating scale-invariant
 # matches that are not contextually appropriate.
 element.locator.found.matches.dimension.deviation.ratio=0.3