gate

Add a semantic gate: a yes/no question a decision model answers about the call right before the function runs. Every gate must pass. When one fails, the data_map doesn’t run, and the AI gets the on_fail output of the first failing gate.

Call gate() once per gate, up to 8. The gates are checked when to_swaig_function() builds the function, which raises ValueError for any gate the platform would refuse. To set what the AI says during the check, use gate_fillers().

Parameters

SemanticGate | dict[str, Any]Required

A SemanticGate, or a gate dict in the platform’s shape (question, threshold, on_fail, and optional id, criteria, visual_input).

Returns

DataMap — Self for method chaining.

Examples

from signalwire import DataMap, FunctionResult, SemanticGate
cancel_order = (
    DataMap("cancel_order")
    .purpose("Cancel one of the caller's open orders")
    .parameter("order_id", "string", "The order number", required=True)
    .gate(
        SemanticGate(
            "Has the caller asked in `conversation` to cancel the order in `proposed_function`?",
            threshold=0.9,
            on_fail="cancel_order was not run. Ask the caller which order to cancel.",
        )
    )
    .webhook("POST", "https://api.example.com/orders/${enc:args.order_id}/cancel")
    .output(FunctionResult("The order is cancelled. Tell the caller."))
)
print(cancel_order.to_swaig_function())