LangGraph Functional API Skill

Expert assistance for the LangGraph Functional API: imperative, Python-native agent workflows using @entrypoint and @task decorators. Gets you LangGraph persistence, streaming, and human-in-the-loop with minimal boilerplate — especially useful for wrapping existing agents or frameworks.

When to choose Functional API over Graph API:

• You're wrapping an existing agent (CrewAI, AutoGen, Strands, etc.)
• Your workflow is naturally imperative (sequential Python logic with branches)
• You need parallel task execution without defining a graph topology
• You want to add persistence/streaming to an agent you didn't write in LangGraph

Full corpus: ../langgraph/references/llms-txt.md (5.4 MB) and ../../langchain-deepagents/references/llms-full.md (10 MB).

When to Use This Skill