feat: LangGraph multi-node research agent on Crusoe Managed Inference

[Sakshi3027]

What this adds

A LangGraph integration example for Crusoe Managed Inference.

Implements a 3-node research pipeline:
Research → Analysis → Summarize

Each node calls ChatCrusoe independently. LangGraph manages state between nodes.

Why it's useful

The solutions library already has a LangChain integration (langchain-crusoe). This extends that to agentic multi-step workflows using LangGraph a common pattern for teams building research agents, RAG pipelines, and autonomous systems on top of Crusoe.

Testing

Tested locally using Groq as a drop-in replacement (same OpenAI-compatible interface). The agent automatically falls back to Groq if CRUSOE_API_KEY is not set, making it easy for contributors to test without a Crusoe account.

To run on Crusoe:
export CRUSOE_API_KEY="your-api-key"
python examples/research_agent.py

Files

• agent.py — core graph with 3 nodes and shared state
• examples/research_agent.py — runnable example
• requirements.txt — dependencies
• README.md — setup and usage guide