Interface NodeLlmConfig

Optional per-node LLM override attached during planning or compilation.

The runtime may use this to route different graph nodes to different providers/models without changing the graph-level default LLM config.

interface NodeLlmConfig {
    providerId: string;
    model: string;
    reason?: string;
}

Properties

providerId: string

Logical provider identifier selected for this node (e.g. openai, anthropic, groq).

model: string

Model identifier selected for this node.

reason?: string

Human-readable explanation for audit/debugging.