Plenty of adventures modules suggest having a well-rounded party. It’s not uncommon for players to ask each other what character types they’re playing to limit if not avoid redundancy. Wizards of the Coast reworked D&D’s mechanics in 3rd edition to make rogues and clerics more fun because of common beliefs that every group had to have one of each even though nobody wanted to play them. I think that way of thinking is all wrong.
Now, to be clear, it certainly tends to be helpful for a party to have as much class variety as possible. Using tools meant for a specific purpose usually works better than using a more general tool. However, one of the great things about TTRPGs is being able to solve problems with creative thinking instead of being limited to a menu of rigidly-defined abilities. I’ll go over ways of enabling creative problem-solving on a conceptual level first, and then I’ll talk about the specific cases of dealing with lacking thief skills and/or spellcasting.
Enabling creative problem-solving sounds fancy, but it’s actually pretty simple in practice, and like some of the points in earlier posts, it’s something that’s mostly about following general good practices rather than anything specific to DM+1 play. The key to it is to avoid using problems for which there are only a limited and finite number of solutions (typically one). Once you have the mindset that there are potential solutions that you don’t know in advance, you force yourself to engage with what the player is trying to do. That doesn’t mean everything the player tries to do should have a chance of working; it means you should listen to what the player says they want to do and act in good faith to adjudicate what happens next. It should fail if it has no chance of succeeding, but as I’d said previously about pacing, failing forward or succeeding at a cost is generally better than simply failing.
Now, there are other tips that can help to enable creative problem-solving.
One of the best pithy pieces of advice is Justin Alexander’s “don’t prep plots, prep situations”, and its smaller-scale cousin “prepare problems, not puzzles” (I can't recall where I saw this before, and a quick search didn't yield relevant results). When you have a toolbox of acting bits and pieces instead of a preplanned sequence of events, it becomes easier to react to what the player is doing without snapping to deciding it doesn’t work.
Another point is to pay attention to how you’re presenting the problem. For instance, in the Wandering DMs’ recent video playing through a DM+1 module, they criticized having about half of the dungeon hidden behind a secret door despite Paul figuring it out without too much trouble, and Dan mentioned having had players fail to find it despite the clue given in the previous room. I think the main problem with the door is the combination of finding it being critical to make progress and not having enough clues to its existence. With a more robust design (such as suggested in Justin Alexander’s “Three Clue Rule” post, and if I seem to be referencing him a lot here, it’s only because his many posts related to running investigation scenarios without railroading are excellent examples of enabling creative problem-solving), that wouldn’t be such a flawed design.
Finally, if there are NPCs around, have them react to what the PC does. This takes some care to avoid giving a clear solution to the player (as it can be difficult for the mind to differentiate between the solution coming from a flawed NPC or from the omniscient GM), but as mentioned previously, having them ask questions, point out oversights, or engage in some Socratic debate can give the player that element of collaboration to spur creativity by nudging them towards lateral thinking.
That’s all well and good in theory, but what about dealing with things that certain characters simply can’t do, like thief skills or spellcasting? Simply providing NPCs with those abilities is a patch that doesn’t address the actual problem, so I won’t discuss that.
Traditionally, thief skills cover picking pockets, climbing, being stealthy, opening locks, disarming traps, and hearing noises. There are two ways to handle them.
My preference is to simply get rid of thieves. I don’t think that they add anything to D&D, their presence detracts from what everyone else can do, and by most rulesets that I’ve seen, their skills represent mundane acts. Just figure out a way to let everyone do those things, and the issue disappears.
If you want to keep them, though, I think the next best option is to still let everyone try those acts with lower chances of success. The exact specifics of how that’s done should be tuned to your specific game, but common solutions are using a lower-level thief’s skills or using a flat X-in-6 (or 10, 12, 20, or whatever other dice) check. I think this works best when playing with a ruleset where an actual thief’s skills are supernatural, and the non-thief's chances be substantially lower than an actual thief's would be. If I was doing this, I'd be tempted to allow a thief to check against the mundane chance of success if they fail their check, too, but that might be getting to the point where there's too much rolling going on.
Spellcasting is trickier, since it’s explicitly supernatural and specialized. For this, though, there is the benefit of having more flexibility to combine multiple options.
The most obvious choice is to let any PC use some magic, even if they normally can’t. I wouldn’t do this right off the bat for a non-caster, but depending on the setting, I’m not opposed to characters gaining limited access to basic spells if they spent the time to learn it. I’m especially amenable to characters gaining some magic by making deals with spirits/faeries/devils/etc., since that’s both such a classic trope and often a great hook for future adventures. Consuming the remains of a magical creature is also a classic set-up for this.
One step removed from that is giving the PC more magic items. Since there’s no need to worry about imbalance between PCs in a DM+1 game, you can be generous with magic items (in quantity and quality) without concerns about marginalizing other characters. That said, I’m not a big fan of this. Maybe it’s my general preference for low fantasy, but I dislike the feeling that the items are overshadowing the character, which can be a pitfall with this. My advice would be to favor limited-use items and/or items with drawbacks (sentient items with strong wills of their own, like Elric’s Stormbringer, being a classic example). You want the item to be part of the character, not the character to become a bearer for the item.
Tying back into the previous section, don’t present problems that need to be overcome by magic. If the problem requires spellcasting, especially if it requires a specific spell, it’s probably not a good problem. There are exceptions to that, but that’s a good general rule of thumb.
Lastly, there’s nothing wrong with dicing for the occasional extraordinary feat. Remember, while he’s supposed to be an all-natural human, Conan managed to break free of the sorcerer’s mesmerism in The People of the Black Circle. The odds should be slim, and there should be a chance of it backfiring to discourage relying on this, but there are countless tales of fate, fortune, and deities showing favor in a singular moment. Obviously, this should be a rare last gasp, not a regular feature, otherwise it stops being special.
Going back to what I’d said at the beginning, I’m not aware of any D&D or derivative ruleset that requires using a variety of classes, even the explicitly party-oriented 4th edition. Thus, it should be possible to play with a party where everyone uses the same class, and taking that situation to its logical extreme is what you get when discussing power variety in DM+1 play. In my experience, the “prepare problems, not puzzles” maxim alleviates most issues with designing custom content, and very few good adventures include obstacles with limited solution sets. Work with your player’s creativity, handle edge cases if and when they come up, and most importantly, have fun.