LangGraph

Not available in the .NET SDK. There is no LangGraph adapter in conductor-ai. This page exists to keep the documentation structure aligned with the sibling SDKs; see the Python SDK LangGraph guide if you need it there.

What to use instead

LangGraph models an agent as a state graph with explicit nodes and edges. The native agent API expresses the same shape:

LangGraph concept	.NET equivalent
Node	An Agent in the Agents list
Edge	AllowedTransitions entry
Conditional edge	OnCondition handoff, or Strategy.Router with a classifier
Cycle / loop	Strategy.RoundRobin or Strategy.Swarm with MaxTurns
Terminal state	Termination condition, or a TextGate on a pipeline
Checkpointer	Durable executions — every run is resumable via ResumeAsync

var team = new Agent("graph")
{
    Agents   = [planner, worker, reviewer],
    Strategy = Strategy.RoundRobin,
    MaxTurns = 6,
    AllowedTransitions = new()
    {
        ["planner"]  = ["worker"],
        ["worker"]   = ["reviewer"],
        ["reviewer"] = ["worker"],     // cycle back on rejection
    },
};

See ../concepts/multi-agent.md for the full transition model and ../concepts/termination.md for stop conditions.

Determinism

Where LangGraph relies on your process to hold graph state, Conductor persists it server-side, so a run survives process restarts without a checkpointer. For a plan executed deterministically rather than decided turn-by-turn, see Strategy.PlanExecute in ../concepts/deploy-serve-run.md.