Node that makes LLM completion calls. More...

#include <node.h>

Inheritance diagram for neograph::graph::LLMCallNode:

Collaboration diagram for neograph::graph::LLMCallNode:

Public Member Functions
std::string	get_name () const override
	Get the node's unique name within the graph.

bool	has_provider () const noexcept
	False when built from a NodeContext without a provider; run() then fails.

	LLMCallNode (const std::string &name, const NodeContext &ctx)
	Construct an LLM call node.

asio::awaitable< NodeOutput >	run (NodeInput in) override
	Dispatch the owned typed request and retain every ordered returned message.

Detailed Description

Node that makes LLM completion calls.

Reads the "messages" channel, builds completion parameters, calls the LLM provider, and writes the response back to "messages". Mirrors the Agent::run() LLM call logic.

Definition at line 198 of file node.h.

Constructor & Destructor Documentation

◆ LLMCallNode()

neograph::graph::LLMCallNode::LLMCallNode	(	const std::string &	name,
		const NodeContext &	ctx
	)

Construct an LLM call node.

Parameters

name	Unique node name within the graph.
ctx	Node context providing the LLM provider, tools, model, and instructions.

Member Function Documentation

◆ get_name()

std::string neograph::graph::LLMCallNode::get_name ( ) const

inlineoverridevirtual

Get the node's unique name within the graph.

Returns: Node name string.

Implements neograph::graph::GraphNode.

Definition at line 211 of file node.h.

◆ run()

asio::awaitable< NodeOutput > neograph::graph::LLMCallNode::run ( NodeInput in )

overridevirtual

Dispatch the owned typed request and retain every ordered returned message.

Streaming observers receive complete typed events; LLM_TOKEN is display only.