langgraph-parallel

npx skills add majiayu000/claude-skill-registry --skill langgraph-parallel --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

# LangGraph Parallel Execution Run independent nodes concurrently for performance. ## When to Use - Independent agents can run together - Map-reduce over task lists - Scatter-gather patterns - Performance optimization ## Fan-Out/Fan-In Pattern ```python from langgraph.graph import StateGraph def fan_out(state): """Split work into parallel tasks.""" state["tasks"] = [{"id": 1}, {"id": 2}, {"id": 3}] return state def worker(state): """Process one task.""" task = state["current_task"] result = process(task) return {"results": [result]} def fan_in(state): """Combine parallel results.""" combined = aggregate(state["results"]) return {"final": combined} workflow = StateGraph(State) workflow.add_node("fan_out", fan_out) workflow.add_node("worker", worker) workflow.add_node("fan_in", fan_in) workflow.add_edge("fan_out", "worker") workflow.add_edge("worker", "fan_in") # Waits for all workers ``` ## Using Send API ```python from langgraph.constants import Send def router(state): """Route to multiple workers in parallel.""" return [ Send("worker", {"task": task}) for task in state["tasks"] ] workflow.add_conditional_edges("router", router) ``` ## Parallel Agent Analysis ```python from typing