RAG reusability

From the related issues listed bellow, and the [documentation about agents](https://github.com/sigoden/llm-functions/blob/main/docs/agent.md#documents) I understand that the files for the agent RAG should be local to the directory that contains the agent's `index.yaml`. With that clear, what is the suggested approach to use an agent with different RAG contents? I am going to give an example of a possible usecase: Let's say that I create a code agent focused on a specific programming language, I have several projects I have made with that language (lets say 50 to say a number). I want to use the code agent on all of them. The agent should be able to create, delete or patch code files in the current project. Because I can't reuse the same agent with different RAG's, Should I create 50 agent definitions where the git directory of each project lives inside the agent directory where the `index.yaml` file is located? If I am wrong, I would appreciate if you can illuminate the right approach for this usecase Related Issues: - https://github.com/sigoden/llm-functions/issues/126 - https://github.com/sigoden/llm-functions/issues/189