Design Notes: How to Write a Spell (Patreon)

I've written a lot of spells. By a quick count (probably off by a few), I'm at about 115 spells personally, with another 159 co-authored with Omega Ankh for Spells That Don't Suck. I've threatened a few times to write a comprehensive guide to authoring 5E spells — and now's the time to give it a shot.

Good lord, this post got so long it needs a table of contents. First, I'll give the quick TL;DR "how to write a spell". Then, I'll go in detail through a bunch of important topics to consider. Last, I'll go through an example spell-writing process.

TL;DR — Write a Spell Step-by-Step

If you just want a quick how-to, here's a 12-step process to follow.

Choose the name

Decide the basic effects

Guess at the damage

Apply any added effects

Choose an "At Higher Levels" effect, if any

Check comparables

Adjust if needed

Repeat 1-7 until you're satisfied

Workshop

Repeat 1-9 until you're satisfied

Playtest

Repeat 1-11 until you're satisfied

Why Do You Need a New Spell?

Step 1 is to decide what to write and why you need it. If you're just attaching spells to a magic item or padding out a spell list, is there something that already does what you need? Between the SRD, Kibbles' Spell Compendium, Spells That Don't Suck, and So Many Spells (all available under Creative Commons, and so very easy to reference), you've got close to a thousand spells to pull from.

Sometimes you do need a new spell, though. The best entry points I usually find are:

Filling out a subclass: You're writing a subclass's spell list and there's nothing quiiiite right. Maybe your theme's really specific, maybe you have a specific mechanical purpose in mind, or maybe there's a "right" spell but it's at the wrong level, or poorly-balanced, or just not quite what you need.

A clear mechanical purpose: You need the spell to satisfy a specific mechanical niche. For instance, KibblesTasty started off his casting compendium by writing expanded elemental-spell sets for every element. The SRD provides quite a lot of coverage for fire spells, but characters themed on non-fire elements really struggle (there aren't even five acid spells in the SRD). Or when I'm working on the necromancer, it turns out that it needs a lot of spells to fill out a proper spell list.

An evocative name: Sometimes all you have to work with is a name. As I'm filling out the necromancer's spell list, for instance, I've got a giant list of brainstormed spell names to use, and I'm working from the names forward to figuring out what type of spells they're going to be.

The Damage Curve

Okay, you're writing a new spell. Does it deal damage? No? Cool, that makes your job harder — we'll get to that later. If it does deal damage, allow me to present the (flawed) spell-design bible.

WotC's Dungeon Master's Guide presents a little guidance for creating new spells, in the form of the text and table above. If you start off by following this table, you won't do terribly, but it's important to note a couple of caveats.

First, fireball exists. Wizards has admitted that fireball is intentionally unbalanced, dealing 8d6 damage as a 3rd-level spell; i.e., on the table above, it's two levels too strong. This winds up skewing damage numbers for every spell in the 3rd-5th level range, because a spell has to compete with fireball in order to be useful. You don't want to go higher than fireball (if fireball is overpowered, than stronger-than-that would obviously be overpowered). But if you write a 3rd-level area spell that simply does 6d6 damage, you can have fun watching your players pick fireball instead, every time.

Second, high-level spells get increasingly nutty. Meteor swarm, for example, is a 9th-level spell that hits a colossal area for 40d6 damage, where the damage curve would call for 14d6. Power word kill will melt any creature below 100 hit points, where the curve only goes up to 82.5 damage. So just know that as your spell level gets higher, following the table will be increasingly unhelpful. But it's still a good starting point.

How do you actually set your damage, then? I tend to think of things as a percentage of the damage curve; i.e. fireball comes in at 133% of the damage curve for its level, averaging 28 damage vs. the expected 21. As WotC suggests in the paragraph above the table, you can bump the damage up by 25% if the spell does no damage on a successful save. You can choose to apply a small penalty for an attack-roll spell (it's usually easier to hit AC than for creatures to fail saves, and you also get the possibility of a critical hit). I typically don't, though — it's not a big enough factor to be worth counting. If you're writing a damage-over-time spell, you should generally aim for it to be a little bit better than break-even after 2 rounds, maybe 120% of the curve or so. If the spell does nothing but damage, it's fine for it to come in a bit above the curve; if it has strong additional effects, it should probably be below it. If the spell targets a very strong save (Constitution) it can come in a bit higher; if it targets a very weak save (Intelligence or Charisma) its damage should get a penalty. And again, keep in mind that the damage curve is wonky in levels 3-5 and 7+.

On healing spells: the DMG includes another paragraph, not pictured, that says healing spells can follow the same damage, and that cantrips shouldn't ever heal. This is mostly wrong, for a simple reason: healing spells can't miss. So if you take a damage spell and haircut its damage by the chance that it misses, you get somewhere between 65 and 82% of its stated damage, usually. (65% is our rule of thumb for a baseline chance to succeed; 82% accounts for a spell that still does half damage on a successful save). Now, 5E also doesn't typically make healing spells as strong as offensive spells, which is a whole tangent we don't need to go on here. But the upshot is that your healing spells should get balanced around 65% of the damage curve.

Lastly, a quick note on 2024E: Wizards actually adjusted the damage curve up for 2024E. (This is a little odd, given the claim that spells port between the editions seamlessly, but the table's only a bit different). It's honestly not different enough to be worth caring about, though, and they didn't follow it themselves for the spells they reworked in 2024E.

Effects & Riders

Whether your spell does damage or not, it's probably going to have some effects beyond just straight damage — otherwise it'd be pretty boring. If your spell is primarily about its non-damaging effects, consider stripping the damage off entirely; mediocre damage can be a red herring for players, causing them to use the spell inappropriately and be disappointed as a result.

A note on "virtual damage": I've seen people try to reverse-engineer a virtual damage metric, trying to assign numerical values to non-damaging effects. (For example, if you consider one round's worth of hard control to be 15 virtual damage, then you can say "hold person deals 15 virtual damage per round".) I don't love that approach. It can get you into the ballpark, but it can't take into account just how effective a control spell can be. Even if you just factor in hold person's ability to prevent the enemy from dealing damage, it's only going to be worth 15 DPR against a fairly low-level monster. Use it against a high-level solo enemy, and you might be preventing 100 DPR while enabling your allies to hit and auto-crit for hundreds more. If you're still only getting into the ballpark, I think virtual damage is an overcomplication.

So how should you rate non-damaging effects? Comparable spells, basically. I'll talk more about comparables later, but think about what other spells apply similar effects. Consider Hold person again, and its higher-level cousin hold monster. Hold person applies the paralyzed condition at level 2, against a limited range of targets (humanoids), and requires concentration. Hold monster relieves most of that target restriction at 5th level. So you can easily say that applying the paralyzed condition with a level 1 spell is going to be too much, and the restrictions are pretty important until at least level 5.

Often, I find it's helpful to dream up an effect for my spell, rough-draft it in, and then check comparables. If I write a spell named paralyze anything, and set it at 5th level, then checking my comparables quickly tells me "I've made this much too strong". So then I can either increase the level or tone down the effects until it starts to look reasonable.

At Higher Levels

Many spells come with an "At Higher Levels" effect (or as I call it, an AHL). Consider adding an AHL if your spell is low-level (to help keep it relevant), if it's intended for Warlocks (since they always upcast things), or if it's easy to add an AHL (simple damage scaling, usually). Consider intentionally avoiding an AHL if it's tough to write (it's a weird effect to scale) or the spell's already very high-level (it's already unlikely to get upcast).

The simplest kind of AHL is just scaling the spell's damage up, typically by one die per spell level. Some control spells can get added targets or larger areas. Some utility spells can get a duration or range increase. For some spells, scaling the damage every level can be too much, in which case you can scale it by every two levels.

Schools, Tags, and Families

You probably want to fit your spell into the spell corpus in a sensible place. Obviously, the spell needs to belong to a school, and that has a mechanical impact. A few important factors to pay attention to:

Abjuration spells are protective in nature, often providing defensive boosts or creating arcane barriers. If you can equate their damage-prevention to a hit point number, you can often weigh them on the normal damage curve (as if they were healing spells). They almost never do direct damage and almost never target a hostile creature. 2024E reclassifies healing spells to abjuration, which is a choice.

Conjuration spells summon creatures, create objects, or manipulate planes or portals. (This last one feels like an odd fit, but in prior editions, summoning was more clearly "calling a creature off of another plane"). A decent number of damage spells are conjuration if they create damaging objects; acid splash creates a big glob of acid on your enemy. These are usually physical effects, so if a spell calls for a saving throw, it's likely Strength, Dexterity, or Constitution.

Divination spells, well, divine things. They detect things, search for things, or reveal things. They should ~never deal damage; this is a pretty important rule entirely because of how WotC wrote the School of Divination wizard subclass. (The one time WotC wrote a damage-dealing divination spell, they intentionally made it horribly weak, to account for that subclass.) In the rare cases they call for a saving throw, it's usually Wisdom.

Enchantment spells apply persistent effects to creatures or objects, and rarely deal damage. Basically every charm spell is enchantment, as well as a bunch of direct control spells and a few buff and debuff spells. In prior editions, this school was a bit broader, but 5E's version is pretty narrow. When they call for a save, it's usually a Wisdom save.

Evocation spells manipulate energy directly. These are almost all direct-damage spells, aside from a few oddballs like sending. And all the oddballs could honestly belong to a different school without much headache. These have the highest share of attack-roll spells, and when they call for a save, it's most often Dexterity or Constitution. Healing spells in 2014, weirdly, are evocation.

Illusion spells create illusions! Any spell with effects that only apply to a creature's senses is generally going to be illusion; sometimes the line between illusion and enchantment can get a bit blurry. Most illusions create visual or audio effects, and they almost always present the semblance of a phenomenon that isn't really there. Some of them can deal significant damage, though; when they do, it's usually psychic damage and usually tied to an Intelligence save. Illusion spells sometimes have a "disbelieve" mechanic, where a creature can break free of the effect if they physically inspect it; this often-but-not-always requires an Investigation check or an Intelligence save.

Necromancy spells are creepy deathy magic. Wilting plants, sapping life force, directly applying damage. When a necromancy spell calls for a save, it's almost always Constitution. Raising the dead is also typically necromancy.

Transmutation spells can be broadly defined as "spells that do stuff, but not any of that other stuff". Spells that change the state of the world without falling into one of the other schools. It's the biggest category and a bit of a catch-all, without clear trends as a result. You can still spot some clear groupings, though. Most shapechange spells are here, as well as spells that move or change existing objects. Lots of buff and utility spells are in this school.

Schools aren't the only way to classify spells, though! You can also think about spell tags. While WotC didn't expressly define any tags in its books, they're a common way to group related spells and key other mechanics off of them. Songs of the Spellbound Sea has ferromancy spells, for instance, which are all metal magic. Blood magic is a common spell tag (though most creators mean something slightly different when they use it). My spells include the osteomancy (bone magic) and shadow magic tags. The "tag" concept lets you hang a mechanic off of a spell tag, creating consistent rules for how a character might interact with those spells.

Last, you can think about effect groupings or spell families. These aren't mechanically explicit, usually, but are apparent when you study the spell corpus. Consider the Tasha's Summons (Summon Animal, Summon Undead, etc.). These are all mechanically similar, offering different spins on the same idea. I already reference the "Hold X" line of spells up above. Wall spells can be a family, but they can also be three separate families: hard walls, soft walls, and semisoft walls. When you're writing a new spell, you can think about expanding an existing family, or write a spell and then fit it back into a family, or ignore the concept. But it's really useful to think about spell families, because it makes it much easier for DMs to understand edge cases in how your spell should work.

Riders and Effect Types

A "rider" is an effect a spell applies on top of whatever its main effect is (typically damage). WotC's spells wave in the direction of spell riders, and if you include Kibbles' and my spells at your table, you'll get a much clearer trend in them. To go over some common ones (mostly codified by Kibbles):

Acid spells often use d4's for damage, and often apply damage-over-time or extra damage at the end of the target's next turn.

Cold spells almost always use d8's and often apply speed penalties or the restrained condition.

Earth spells use d10's and often knock creatures prone or shove them.

Fire spells use d6's for damage, and generally have high damage. Their special feature is "do more damage", usually. Occasionally they set things on fire, but not often; it's hard to keep that balanced.

Force spells often use d4's and deal automatic damage (no save or attack roll required). While that often sounds overpowered, their damage is typically low enough that it's not.

Lightning spells use d12's and can shock or stun targets. Sometimes they jump between multiple targets.

Poison spells usually use d12's and can poison targets (duh).

Thunder spells usually use d8's and often inflict knockback or deafness.

Water spells usually inflict bludgeoning damage and can shove targets or knock them prone.

Wind spells often inflict slashing damage and use d8's.

Those are KibblesTasty's categories, based on the spells WotC wrote but codifying the rules a bit more tightly. A few purer-homebrew categories below:

Blood spells cost the caster some hit points to use, but often have higher damage and avoid concentration requirements. They almost always deal necrotic damage, and many of them will drain some hit points from targets.

Bone (osteomancy) spells often deal piercing or bludgeoning damage, sometimes combined with necrotic. They often impose healing penalties on the target.

Ferromancy spells usually deal piercing or slashing damage, and often impale targets. They may have different effects based on a secondary element or based on the type of metal they evoke.

Shadow spells usually deal necrotic damage, and often interact with light and darkness.

Other creators might have different definitions, but this is where I start from! In general, spells with stronger riders (e.g. poison or restrained) should have lower damage or target stronger saves; spells with high damage or targeting weak saves should only have weak riders.

Comparables

I mentioned comparables in a couple spots; the concept here is pretty simple. You should benchmark a spell by whatever the closest existing spells do. To pick on fireball again: a mid-level area-damage spell can treat fireball as a comparable. It's a bad idea to be stronger than most important damage spell in the game, but you can't afford to be too much weaker. You're both looking to make sure your spell makes sense in the context of the spell corpus, and making sure players won't just choose other existing spells over yours.

For all spells, your final step (bar playtesting) should be checking it against a few comparables at the same or adjacent levels, and making sure you haven't overlooked anything. For non-damaging spells, where your starting point can't be the damage curve — you might check comparables for both your first and your last steps.

Tiering & World Fabric

There's a concept I call "Big Magic". Spells that just intrinsically are powerful, even if their mechanical impacts on the game aren't. Teleportation is big magic; only powerful spellcasters get to teleport. The game's basic teleportation spell is teleportation circle, which is available at 5th level and has significant restrictions. Every other long-distance teleport is higher-level. Only high-level spellcasters can do long-distance teleportation; that's a rule of the world, reinforced by the spell corpus.

Broadly, when you write a spell at a given level, you should think about what sorts of spellcasters can use it. If it's a cantrip, then every apprentice and acolyte can do it. Let's invent a cantrip (say, Radagast's Runic Rhûmba) that sweeps up dust and mops floors; that's the only thing it can do. Is this mechanically balanced? Sure, you won't slay any villains by cleaning the house. But does it change the world fabric? Substantially. Any apprentice wizard can start a wildly-successful house-cleaning business. Every castle and temple would have its own janitorial sorcerer.

Similarly, If you let prestidigitation perfectly clean clothes, you'll have a wizard laundromat on every corner. Zone of truth is 2nd level, and is honestly too available; the ability to compel absolutely truthful answers would have huge impacts on any nation's system of justice. Hell, if you let guidance apply to speech checks, every ruler should have a temple acolyte at their elbow at all times.

The point of this is to think about the effect your spell will have on the world, mostly for utility spells. If it's a cantrip, you want something that can't change the way the world works even if ~everyone has access to it. For 1st- and 2nd-level spells, you should expect those to be pretty readily available for people with sufficient resources. 3rd- and 4th-level spells start to get more rarified, but it's still not hard to imagine a powerful figure keeping a 7th-level spellcaster handy. 5th level and up is where Big Magic starts to kick in; spells here can have potentially world-altering effects. The key is that any caster able to use such high-level magic has better things to do with their time (and spell slots) than just operating the magic subway.

To Concentration or Not To Concentration?

Concentration is the biggest balance lever spells have. It's a solid mechanic and a good answer 5E deploys to the buffing problem. You can summarize the buffing problem as "the key aspect of high-level play in earlier editions". Namely, in 2E or 3.5E or Pathfinder, the way to handle key combats is to start by casting all your buff spells at once; potentially layering a dozen or more buffs onto a single super-soldier. This is sort of fun the first time you do it, very tedious every time after that, and a giant headache for your DM every time. 5E's buff spells are (almost) all concentration, which greatly limits this type of shenanigan.

In general, concentration's appropriate for any spell that has an effect over time, whether that's a persistent effect, damage over time, a long-lived buff, or something else. But a non-obvious and huge consideration with concentration spells is how they play out differently across the spell levels.

For cantrips, concentration is an enormous penalty. It's not always inappropriate, but a caster will typically want to spend their concentration on a leveled spell, so it's just a huge hit to a cantrip's usability.

For 1st-3rd level spells, concentration works great. You'll make a big choice about which spell to concentrate on, based on which effect the party most needs.

For 4th+ level spells, concentration starts getting thorny again. The issue is that by this point, you've got a ton of good options to spend your concentration on, including lower-level ones. Any new spell that requires concentration is now competing with haste, polymorph, and other all-time bangers, and as you reach higher spell levels the competition gets tighter.

Spell Lists

Once you write a spell, you need to decide who to give it to! This, again, can be trickier than it seems.

Bards get poor damage options, but a ton of control and utility magic. Lots of charms and mind-altering stuff. A little healing, not a ton.

Clerics get sub-par damage options, but lots of buffs, some utility, and basically every healing spell.

Druids get healing spells, lots of utility, and underwhelming damage options. Their spell list (especially for damage) is also heavily dependent on concentration spells.

Sorcerers get basically the exact same list as wizards, but they can get a few goodies that are "pure" magic. Stuff like wild magic spells, etc. This represents their stronger, more innate connection to their magic.

Warlocks get every "creepy" spell. They get good damage, some utility, some buffs, and lots of debuffs. Since their spells automatically upcast as they level, any spell without a decent AHL (or an effect that's always valuable) gets left behind very fast. (Note that giving warlocks every creepy spell is a WotC default; as I've written more necromancer spells, I explicitly avoid giving them all to warlocks, to help differentiate the two classes.)

Wizards have the best spell list in the game, with the most spells and the largest variety. About the only thing they struggle for is healing magic.

Rangers and Paladins get spell lists that mostly look like their full-caster counterparts. Both of them do get some unique goodies though; rangers get more combat tools, paladins get all the smite spells, plus a few more.

For my own classes:

Troubadours are a bardic half-caster and get a similar spell list to bards. Damage options are generally limited to thunder, with a little psychic, and aren't great. They get lots of buffs, debuffs, and soft control spells.

Bloodragers are a sorcerous half-caster, but with tools for boosting their spell damage. Accordingly, they get a lot of damage spells, particularly elemental ones. They also get a decent amount of iconic sorcerer spells, just to feel more sorcerous.

Necromancers get most necromancy (duh) spells, some transmutation, a little enchantment and illusion. Damage spells are generally limited to necrotic, poison, and cold. Decent utility options but not top-tier. They're very concentration-heavy, similar to druids.

I tend to find spell list assignment is the thing that changes the most as you add more homebrew classes to your game. After all, if I add a cool necromancer spell to a game with no necromancers in it, somebody should get it. I also think this is one of the places we all tend to mess up our home games a little bit — we always grant more spells to classes, but very rarely take them away. Trimming spell lists sometimes would help keep class identities more distinct.

Pitfalls

There are some clear problems we can identify, things that crop up repeatedly across poorly-balanced spells.

First, the slot-efficient trap. Consider witch bolt. For one 1st-level slot, you get to do 1d12 automatic damage to a target every round, no save, no attack roll. In theory, if you maintain it for 10 rounds, that's 10d12 damage. 8th-level value for a 1st-level spell slot! But the spell's crap. It's got some big usability problems, but even once you get past those, it's not time-efficient or action-efficient enough to be good. Immolation is another good illustration; a 5th-level spell that deals free damage to the target every round. If the target fails every save, and you maintain concentration for a minute, you're doing 154 (44d6) damage for a single action, which is insane. But in practice it's very bad, because that damage takes so long to come out and you can be spending your concentration for more efficient spells.

Second, the double-failure rule. Consider ray of sickness. A 1st-level spell that makes an attack roll, deals low damage, and then poisons the enemy on a failed save. If the enemy makes their save, the spell underperforms quite a bit relative to other 1st-level spells. So you're testing two defenses at once, and you have to get through both in order to get good value for your spell. If either defense (the enemy's AC or saving throw) succeeds, your spell isn't very effective. Still, it's super-tempting to write double-failure spells! And it's not always wrong, as long as the spell's value still works out. A ton of effects just mechanically make sense this way.

Third, unbeatable spells. Most of 5E's worst-balanced spells are in this category. Wall of force can remove an enemy from the combat without testing any defenses at all. Telekinesis can restrain an enemy in a way it can only resist by having a high Strength score, even it it's otherwise a super-boss with legendary resistances. Sleep only cares about how many hit points the target currently has; it tests player knowledge rather than any game mechanics. Banishment or hypnotic pattern only test the enemy's saving throw one time before applying a full minute of control. The answer here is simple: don't write spells like this.

Fourth, no-niche spells. These are spells which are fine in a vacuum, but don't actually have any valid purpose that another spell doesn't do better. Damage spells can fall into this trap if they're just underwhelming next to another option. Above, I mentioned a hypothetical 3rd-level spell that does 6d6 damage; it's fine by the damage curve, but pointless when it sits next to fireball. For utility or defensive spells, this can be harder to assess, but the same problem absolutely comes up. Consider divine favor, a 1st-level spell that grants all your attacks +1d4 damage. Sounds great, right? But it sits right next to bless, available on the same classes, that grants three characters a +1d4 bonus to hit; generally twice as valuable as the same damage boost. Divine favor isn't completely pointless; if you have two casters with both available, you can stack them on the same PC for bonus damage and to-hit. But it's so heavily overshadowed by bless that it's very rarely used.

Shut Up You Friggin' Pedant, Write Me a Spell Already

I'll pluck a random spell I've written recently to use for our example. Murder of crows is a 7th-level necromancer spell.

Murder of crows started as a spell name, inspired by the piece of art above. I decided basically from the get-go to aim for a high level spell, because that's where I've got more work to do padding out the necromancer's list. So I looked at the corpus and thought about how I could reflect the theme while getting something mechanically interesting. Note that this is a spell I've already written, so I'm sort of simulating the writing process here.

Choose the name: Done! Easy one here.

Decide the basic effects: The theme of "getting attacked by a flock of crows" works well for a large area of effect, so let's do that. This also gives us a unique way to target the spell — movies like "The Birds" give us a clear idea of an individual person getting swarmed by birds. So let's make it an AoE spell, but one that can target a single creature and move with them. And crows aren't an instantaneous effect, so let's make it persistent; this suggests we should make it concentration too. A Dexterity saving throw makes sense for the flavor, and helps us with mechanical variety, because our other necromancer spells heavily skew towards Constitution.

Guess at the damage: A flock of crows should be inflicting lots of small hits, so d4's are appropriate to represent that. The curve suggests 42/12d6 damage for a 7th-level AoE. But this is a damage-over-time effect, so we should aim above the curve. 120% would be 50 damage, and probably a good place to go per the damage curve. 50 damage in d4 terms is 20d4, so 25/10d4 per round makes sense.

Apply any added effects: Getting attacked by a flock of crows would definitely be chaotic and disorienting, but not in a way that would apply any super-powerful effects. Heavy obscurement and deafness both make a lot of sense. Applying a penalty to concentration spells makes sense and gives this a somewhat unique mechanical slant as a high-level anti-caster spell.

Choose an "At Higher Levels" effect, if any: We're already at 7th level, so there's not a lot of motivation to attach an AHL. If I really wanted to, I could increase the radius or up the damage by 2d4 per slot level.

Check comparables: Sleet storm is an interesting one; much lower level, but one of the few other spells that directly counters concentration. Circle of death (and its SDS equivalent, necrotic sphere) are 6th level and are similar large-AoE damage spells (28/8d6 and 36/8d8, respectively). Whirlwind is the base game's only 7th-level AoE damage-over-time spell, and not from a terribly well-balanced book, so let's not give it too much weight. Abi-Dalzim's horrid wilting is one level higher and deals 54/12d8; that one also definitely belongs on a necromancer list. This is all making me feel like murder of crows might be a little weak at the moment.

Adjust: Let's go up to 30/12d4 damage per round. That's 140% of the damage curve over two rounds; it feels like it's in a good spot for feedback.

Repeat 1-7 until you're satisfied: I'm feeling pretty satisfied.

Workshop: Not everyone has the luxury of a dedicated workshop thread in their discord, but I do! I workshopped this one and it got tentatively positive feedback, with no suggested changes.

Repeat 1-9 until you're satisfied. This one's probably in a playtestable spot (though testing high-level spells is always hard to do).

Playtest. That's the step this spell is at right now! Once I get feedback, I'll do this all over again.

Repeat 1-11 until you're satisfied. :saluting-face-emoji:

For anyone who can't see the image, it landed on:

You summon a massive flock of spectral birds around an empty space or a creature you can see within range, which assails every creature within a 30-foot radius. At the start of each of its turns, a creature within the flock must make a Dexterity saving throw, taking 12d4 piercing damage on a failure or half as much on a success. The flock's area is heavily obscured, and creatures within it are deafened. In addition, creatures within the area make saving throws to maintain concentration with disadvantage, and any creature within its area when the spell is cast must immediately make such a saving throw, with the DC equal to your spell save DC.

If the flock is circling a creature which falls to 0 hit points, you can use your action to designate a new space or creature for it to circle.

Conclusions

I dunno, man, I've been writing this for way too long already. Write spells! It's fun! They're relatively bite-sized, comparatively easy to test, and if you have a good idea you can probably knock one out in half an hour. I'll probably revise this guide somewhat at some point, so feedback is welcome!