Personally, our peeves when it comes to the tropes in fantasy books, TTRPGs, and shows can be reduced to four categories. Our favourite works are those that are careful about all of them.
Category 1. Yes We Are Super Pedantic.
Armour made of interlinked rings has a name. It's 'mail', not 'chain mail'. If you want to be period-appropriate, you might say 'maille' or 'mayle'. Armour made of thin lames of metal has a name. It's 'lamellar armour', not 'splinted mail'. The preposterous place we're in with pervasive fantasy armours that are mis-named or just plain stupid can be traced back to Victorian antiquaries and burgeoning medievalists like Grose and Meyrick. They chose to use 'mail' to mean 'metal armour', contra its historical meaning, and invented a bunch of armours (banded mail, tegulated mail, splinted mail, etc) that are historically unattested!
A gorgon is a mythological monster or demon from ancient Greece with a visage that turned people to stone. Stheno, Euryale, and Medusa were the names of the three gorgons. Medusa, and in later stories all the gorgons, had serpents for hair. Medusa was the only mortal gorgon. Having 'medusas' and then unrelated 'gorgons' is like calling countries 'australias' and then saying that the word 'countries' means 'sports teams'.
Historically, alchemy has been more about enlightenment, spiritual refinement, and purity of the self, than it has been about mystical proto-chemistry.
Necromancy means getting information from the (spirits of the) dead, not animating dead bodies.
...And so on.
Category 2. No Thought Given to Consequences.
'Fantasy' doesn't mean 'anything can happen at any time without underlying rule or predictable reason'. That's a description of a dream sequence, or certain strains of postmodernism. Sometimes we get peeved by encountering medieval facades with no real thought given to the consequences of having magic and monsters.
Even something as seemingly essential as the towering stone castle is the product of social and political forces: castle building is a consequence of a fractured state that lacks strong central coordination, instead delegating power to local or regional leaders willing and able to use violence to maintain control. Those leaders build castles (i.e., fortified residences of local elites often used for administration including taxation) to (generally) shore up their vulnerabilities and (specifically) as a vital defence against their nearby peers, in a context where land is wealth and the central state isn't powerful enough to police them.
Castles exist in those conditions as long as there are the resources to make them, and as long as there aren't technologies that can trivially defeat them. Note that castles were replaced by star fortresses in the era of powerful artillery. Lots of common fantastic elements can defeat stone walls at a distance or under cover of stealth: flight or levitation, rock-to-mud or other transmutations, giants or other really big creatures, instantaneous travel, earthquakes, and powered-up versions of mundane siege attacks, like tunnelling, throwing big rocks, or inflicting plague or demoralisation. Such elements entail fewer castles - or none, if they are sufficiently available. Maybe star forts are the way to go in this particular fantasy setting. Or bunkers. Or enormous moats. Or maybe magic is the only defence that works.
Of course, if fantastic elements make it easier to build castles, that provides an impetus in the opposite direction. If good permanent masonry can be conjured by any town magician, then yes, castles might be everywhere despite being permeable to magic... but also, every town will be furnished with multiple curtain walls, aqueducts, tall stone buildings, stone roads, and so on.
In general, every fantastic element in a setting needs to be considered in terms of: how does this affect the overall milieu? If every other mage can send a magical message a hundred miles and every town has at least a couple of mages, what does that imply for a ruler's ability to receive information, wage war, and administer the state at a distance? Because it's sure not governed by the speed of a messenger on horseback any more.
Category 3. Assumptions Based on Modern Culture.
Writers all too often take for granted that some aspects of a free, capitalist society will just show up in a setting: a cash economy, an absence of individual obligations to people higher in the pecking order (or a lack of a pecking order), freedoms of speech and movement and so on, and a modern sense of morality where things like 'don't torture animals' and 'don't execute prisoners' are obvious rather than a weird joke.
Obviously some of these things are desirable to have in a setting just so that readers or players won't be icked out. And obviously if a game or novel requires its audience to pursue a comprehensive course of historical/sociological/economic study before they can engage with it, that game or novel is limiting itself in a pretty major way. But if a setting has these things, it should have a reason to have them. It's not actually that difficult to do - 'gods literally exist, they desire these things for a reason X, and they have worked to implement them by means Y' gets you most of the way there in one sentence. Our peeve is when the world is 'modern Western civilisation with the serial numbers written in Blackletter' and that's just ...how it is. The average powerless pre-industrial commoner can go where they want, choose where to work, says what they want to who they want, has money and spends it, doesn't live in fear of the powerful, and is a devout utilitarian ethicist. Without any of the necessary structures to enable that.
Category 4. Maladaptive Feedback Loops.
There's something that's happened since the dawn of TTRPGs. Game masters - or players - excise the parts of the rules they don't like or don't see the need for. The least popular rules get ignored, popular house-rules spring up and pass on, and in the worst case, poorly-laid-out rule books result in certain good ideas just never making it to the table. Popular voices spread their own gameplay changes through the hobby - first through magazines, then blogs, now social media - and widespread changes in gameplay result.
Later generations of designers create new editions of old games, or make their own new games, based on the way they remember playing or the way people seem to want to play. The feedback loop completes. Some innovations are for the better, but useful material is also lost.
And so you get big successful games about delving into dungeons with whole books oriented towards game masters which somehow forget to teach said game masters how to properly run a dungeon delve. Or sections on realm management which boil down to 'wing it', where bygone editions had exceptional rules for that baked in. Or wilderness survival rules so brief and generous that choosing the right starting character negates any possible wilderness survival gameplay.
Well, so what?
In our own work, we like to put together material that bears all these things in mind. We said 'peeves', but really, it's about forming a cohesive, interesting whole that won't turn off a reader who knows too much and doesn't expect too much out of a new reader/player who's coming from a different genre, style, or expectation.
