Langchain Agent Scaffold

--- model: claude-sonnet-4-0 --- # LangChain/LangGraph Agent Scaffold Create a production-ready LangChain/LangGraph agent for: $ARGUMENTS Implement a complete agent system including: 1. **Agent Architecture**: - LangGraph state machine - Tool selection logic - Memory management - Context window optimization - Multi-agent coordination 2. **Tool Implementation**: - Custom tool creation - Tool validation - Error handling in tools - Tool composition - Async tool execution 3. **Memory Systems**: - Short-term memory - Long-term storage (vector DB) - Conversation summarization - Entity tracking - Memory retrieval strategies 4. **Prompt Engineering**: - System prompts - Few-shot examples - Chain-of-thought reasoning - Output formatting - Prompt templates 5. **RAG Integration**: - Document loading pipeline - Chunking strategies - Embedding generation - Vector store setup - Retrieval optimization 6. **Production Features**: - Streaming responses - Token counting - Cost tracking - Rate limiting - Fallback strategies 7. **Observability**: - LangSmith integration - Custom callbacks - Performance metrics - Decision tracking - Debug mode Include error handling, testing strategies, and deployment considerations. Use the latest LangChain/LangGraph best practices.