Function Calls

Execute Python functions and save their results to state using @effect func. This avoids consuming LLM tokens for tool calls and lets you invoke deterministic logic directly instead of relying on the model to select the correct tool.

Note: The Python module must be available in the activated Python environment and the import path must be correct.

from my_module import compute

@effect func compute(x) => result

The example above calls compute(x) and stores its return value in the result state variable. You can reference this value in subsequent prompts using ${result}.