nxt::ai::agent namespace

Classes

struct output_item_result: Result of consuming one output item from the event stream.
class response_continuation
struct response_stream_result: Summary gathered while consuming one Responses stream.

void append_stateless_turn(nlohmann::json& input, std::vector<nlohmann::json> output_items, std::vector<nlohmann::json> tool_outputs): Append model output items followed by tool outputs to a stateless input.
auto is_event(const responses::stream_event& event, std::string_view type) → bool: Test whether an SSE event has the requested Responses event type.
auto output_item_from_event(const responses::stream_event& event) → std::optional<nlohmann::json>: Extract the event payload's item object when present.
auto output_item_type(const responses::stream_event& event) → std::string: Return the output item type carried by an event, or an empty string.
void prepare_tool_request(responses::openai_responses_request& request, const std::vector<tools::function_tool>& tool_list): Attach tool definitions and stateless-continuation includes to a request.
void request_encrypted_reasoning(responses::openai_responses_request& request): Ask Responses to include encrypted reasoning content in output items.

Append model output items followed by tool outputs to a stateless input.

Test whether an SSE event has the requested Responses event type.

Extract the event payload's item object when present.

Return the output item type carried by an event, or an empty string.

Attach tool definitions and stateless-continuation includes to a request.

Ask Responses to include encrypted reasoning content in output items.