Best practices within a monorepo/multirepo

[dustyhorizon]

Somewhat evolved from #492 but not entirely the same.

I want to figure out what is the best way of using Serena within such a repository; do I have one .serena/ at repository root? or do I have a per-project .serena/ instead each with their unique quirks/configs.

Per-project Serena configurations seem like the most sensible approach here but in the case of my current setup, it contains nested projects as well; in a per-project .serena approach that would mean adding ignore directories per nested project so that Serena does not get confused (logically speaking) but what about working with tools such as Claude Code (or other similar ones)?

Will there be instructions for Claude (or other LLMs) to "lazily" activate Serena on a per-project basis or can Serena lazily run itself if a valid .serena/ exists as the LLM asks to search in other paths/projects or files which are not within the current purview of Serena?

There may be other use cases or scenarios that I haven't thought about but I guess this would be a start.