adk-langgraph

name	adk-langgraph
description	Orchestrate stateful, multi-agent ADK workflows with LangGraph: graphs, state machines, conditional edges, and LLM-driven routing, using ADK tools as nodes. Use when a request mentions LangGraph, a state-machine/graph workflow, complex branching orchestration, or ADK↔LangGraph interop.

adk-langgraph — LangGraph Orchestration for ADK

You build stateful agent graphs with LangGraph that drive ADK tools/agents.

Doc loading: context7-first, local fallback

LangGraph evolves fast. Load the freshest docs adaptively:

1. If the context7 MCP is available, call it first: resolve-library-id "langgraph" → get-library-docs for the topic at hand (graphs, StateGraph, conditional edges, checkpointing).
2. Otherwise (or to confirm the ADK interop pattern), read the bundled reference: references/langgraph-orchestration.md — the full ADK↔LangGraph integration guide (905 lines: state schema, nodes, edges, conditional routing, ADK tool nodes).

context7 is optional runtime enrichment; the bundled reference is always sufficient.

When to reach for LangGraph (vs native ADK workflows)

• You need an explicit, inspectable state machine with cyclic/branching control flow
• Routing decisions depend on accumulated graph state, not just the last message
• You want LangGraph's checkpointing alongside ADK tools as graph nodes

For simple linear/parallel/loop pipelines, prefer native ADK (adk-agentic-prod-workflows / adk-agents) — don't add LangGraph weight.

Process

1. Load docs (context7 → else references/)
2. Define the state schema and the node set (ADK tools/agents wrapped as nodes)
3. Wire edges + conditional routing; add a terminal condition
4. Validate models (adk-litellm); emit runnable graph code

NEVER

• Use deprecated/blocked models
• Build a graph with no terminal/exit edge