Run git command asynchronously

Disconnect from the database.

Extract the first class name from source code. Args: code: Source code string Returns: F

Get benchmark instance by name.

Dynamically load a class from source code. This function creates a virtual Python module in memory (not on disk) by: 1. Gener

Register a tool class or instance asynchronously. Args: tool: Tool class or instance to register config: Conf

Get the next task to be tested.

Perform a protocol transformation. Args: type: Transformation type (t2e, t2a, e2t, e2a, a2t, a2e) env_names:

Create a prompt instance and store it. Args: prompt_config: Prompt configuration force_rebuild: If True, forc

Create a Pydantic model from a parameter schema. Args: name: Name for the model (will be converted to PascalCase + "Input

Build OpenAI-compatible function-calling representation. Args: name: Name of the tool/action description: Des

Build a human-readable text representation. Args: name: Name of the tool/action description: Description

Clean invalid values (NaN and infinite) from a NumPy array or PyTorch tensor. Args: arr (Union[np.ndarray, torch.Tensor]): Input array or

Create a table. Args: request: Table creation request Returns: Query result

Generate next version number for a component Args: component_type: Type of component (tool, environment, agent)

Get value by id

Get the parameters of a function or class from source code. Args: object (Union[Type['T'], Callable]): The function or cl

Returns a short version of the value of the variable. We sometimes use it during optimization, when we want to see the value of the variable,

Set all update flag of namespace indicating all workers need to reload data from files

Update an existing memory system with new configuration and create a new version Args: memory_name: Name of the memory sy