I'll share my thoughts on the subject as I relate a lot to this question. Recently I've decided to jump on the bandwagon of adult game development and see what I can offer to the genre, along with introducing new and interesting ideas. I started off with a story based approach, mapping out plot arcs, narratives, backstories, and all this content. I got pretty far and began working on the game itself but it dawned on me that I wasn't really doing anything different. Just because I'm telling a different story doesn't mean it's going to be that much of a different game. This troubled me because it was quite difficult to make this game stand out if it's only unique pillar is the story. I would strongly recommend you consider that yourself, if the only unique pillar to your game is the story and not the delivery or content of your game then you may accidentally create a another daz copy-paste.
After this I decided to throw everything out the window and recollect on what I truly wanted to achieve, the goals, and my personal rule set of what I want in the game and what I want to get out of the game and so forth. This led me to a genre which you happen to mention above which is roguelikes. For me it was a clear decision that this was my option and what I was looking for. I wanted a new approach on the deliverance of adult content, I want an interactive and replayable world, and I wanted overlapping systems and fetishes.
Now I reached this conclusion based on examining my own pillars, rules, goals, and vision for the type of game and the feel of the game I want. I think it's important to really focus on how you want the game to feel and play because that's going to dictate how you deliver your content whether it be through a story or other mediums, and then decide what kind of tech you can use.
Tech is a big deal, most small game developers pivot their game around what is available to them and what they are proficient in. If you are a huge writer and not a great designer/programmer then Renpy or the visual novel tech is your home. Maybe you can write a little, do okay renders, and are decent at programming. Well then using a tech like RPG Maker fits your needs. I'm not going to lay out every combination of skill sets, but my point here is that often times the scope and scale of your game is limited to your skills and what your professional and you shouldn't disclude that when choosing the type of game and the tech that you're going to use in it.
For an example, I am a terrible writer and I am terrible at storytelling. However I know a little about game design, and I'm familiar with a game framework which lets me build just about any 2D game I would like. This also gives me a lot of freedom because I know how to program. I also I'm alright at doing renders and this leads me to where I'm at now. I could have been working on that story game that I mentioned above, and struggled every day to write a story that's coherent, empathetic, and interesting, but I I know I wouldn't have enjoyed myself. Instead I'm working on a roguelike that focus heavily on replayable content and and fetish variety. I'm having way more fun now then I did at any given point upon working on that previous project.
Tl;Dr the question your asking is pretty in depth and relies heavily on your personal vision, tech available, your skillset, and who you are. I feel there is a lack of sandboxes, and experimentation on what does and doesn't work in them. But even a sandbox is out of reach if you're limited to renpy and limited programming knowledge. So it all depends really