rag-agent

To scaffold a new RAG agent, provide an agent name, domain description, keywords list, and vector collection name. This generates the agent class, prompt template, and unit tests within a consistent rag/ project layout.

Structuring LangChain RAG agents requires a consistent directory layout like rag/agents, rag/prompts, and rag/tests. This approach enforces separation of specialized agents and provides CI-friendly scaffolding for each domain.

Yes, the RAG agent scaffold generates a test skeleton under rag/tests/unit/. It validates the can_handle and process behaviors to ensure your domain agents function correctly before routing them into the application.

Yes, you can use this scaffold to add domain agents to an existing system. It generates a router entry for the new agent, allowing you to register and enable routing seamlessly within your current application architecture.

Yes, you need a vector collection name to generate the agent. The scaffold links the base agent class directly to your specified vector store collection to enable retrieval-augmented generation for that domain.

Creating separate RAG agents allows for domain-specific prompts and targeted retrieval. This scaffold supports adding new domains and prototyping specialized assistants by generating isolated agent packages with dedicated prompts and tests.