I started with one CLAUDE.md, but as I’m building more things, it makes more sense to separate different types of work instead of trying to make one file cover everything. My setup: UGC Revenue - brand research, scripts, creative strategy, portfolio work, client work, and everything around creating content for brands. Local Business - websites, content, creative strategy, and eventually AI and workflow services for local businesses. What each context includes: UGC Revenue - how I research brands, how I approach scripts and creative, what a finished piece should look like, and the rules I follow when I’m working on client or portfolio content. Local Business - information about the business, what their website or content actually needs to accomplish, how I scope the work, and eventually the systems and AI workflows I build for them. I’m separating them because even though there is some overlap, they’re different kinds of work and Claude needs different context depending on what I’m doing. One fact, one location - I don’t want the same rate, rule, or decision sitting in three different files. If something changes, I want to update it once and know Claude is reading the right version. Clean sessions - if I’m working on UGC, I want that conversation focused on UGC. If I’m working on a local business, I want that project focused on that business instead of dragging unrelated context into it. Team consistency - right now this mostly means keeping the different AI tools I use working from the same information. Later, if I bring more people into the workflow, they should be able to follow the same setup without me having to explain everything from scratch. The first thing I want to test is running a similar research task in both projects and seeing whether the context actually changes the direction Claude takes. If the UGC project thinks more about creative angles and the Local Business project thinks more about the business and what it needs, then the separation is doing what it’s supposed to do.