Nagpa Tactics

Thought I might be able to tackle something easy after the drow matron mother, but no—you guys want me to look at the nagpa, another monster with eleventy billion spells. (OK, it’s got 26. That’s still a lot.)

It’s not like you’ll even find nagpas running around all over the place. According to Mordenkainen’s Tome of Foes, there are only 13 of them—a coterie of conniving wizards cursed by the Raven Queen and turned into skulking vulture-people scavenging the remnants of lost civilizations for scraps of arcane lore.

Unsurprisingly, nagpas’ ability contour is that of a long-range spellcaster, with extraordinary Intelligence and Charisma, exceptional Wisdom and very high Dexterity. They carry staves, which they somehow are able to use as finesse weapons and deal two dice of damage with, but melee engagement really isn’t their style. If they do get into melee, they want to get back out of it quickly.

They have proficiency in all the mental saving throws, but their Dexterity and Constitution save modifiers are unremarkable. Thus, they don’t have a lot to fear from bards, whose spells tend to emphasize enchantment, illusion and crowd control; but casters who sling damaging evocation, transmutation and necromancy spells pose a threat that they need to neutralize quickly. Taking out these foes is even more important to them than taking out melee fighters.

Although their flavor text says they “work in the shadows” and “emerge [only] when they can deliver a finishing blow,” nagpas lack Stealth proficiency, and their Dex isn’t high enough to allow them to hide effectively. They also lack illusion spells, “trap” spells such as glyph of warding, and scrying spells. Ambush, therefore, isn’t a viable attack strategy; any plan they concoct has to work without their targets’ being surprised. They do, however, have proficiency in Deception, as well as charm person, suggestion and dominate person, which they can use to manipulate others into doing their dirty work for them. Afterward, the nagpas can liquidate the now-expendable marks. Their multiple language proficiencies can enhance the believability of their ruses.

While it doesn’t affect their tactics in any meaningful way, nagpas’ combination of Perception proficiency and truesight is potent. They may not be able to stage an ambush of their own, but on the flip side, hiding from one of them is nearly impossible.

Nagpas have two distinctive features: Corruption and Paralysis. Corruption is a bonus action (read: action economy enhancer) that affects a single creature, imposing the charmed condition until the start of the nagpa’s next turn. As debilitating conditions go, charmed is relatively weak in combat—all it really does is prevent the target from attacking the charmer. (It’s much better in social interaction, when a nagpa can use it to lay the groundwork for subsequent Deception.) But it does have the benefit of preventing an opportunity attack, which a nagpa can use against a melee opponent to disengage without having to take the Disengage action—as long as there’s only one melee opponent it has to get away from. And one opponent not attacking you is one opponent not attacking you, so there’s that as well.

Paralysis is the keystone of the nagpa’s kit. It’s also a bonus action, and it can potentially paralyze every creature within 30 feet for up to a full minute. The nagpa’s DC 20 is high enough that even high-level PCs may take a few rounds to shake this effect off. Even so, the time to take advantage of it is immediately. How can a nagpa do the greatest possible damage to a half-dozen paralyzed enemies—who automatically fail Strength and Dexterity saving throws and can be attacked with advantage, for automatic critical hits at close range?

It’s kind of amazing, but among all those 26 spells, there are only three that fill the bill: fireball, wall of fire and prismatic spray. And each of them fills it imperfectly:

  • Fireball seems like a gimme, but it has only a 20-foot radius and thus may miss targets caught in Paralysis’ 30-foot radius. The smaller circle covers less than half the area of the larger one (four-ninths, to be exact); normally we’d expect it to encompass four targets, but in this case, we’re going to say it can be assured of hitting only three of the six.
  • Wall of fire has two options: a straight line or a circle with a 10-foot radius. Against paralyzed foes who don’t get to repeat their saving throws until the ends of their turns, these walls are effectively 10 feet wide, rather than 1 foot (although targets they’re drawn directly over take an additional tranche of damage). The circular wall can therefore be either a disc of flame with a 10-foot radius or a ring of flame with a 20-foot radius but a 10-foot-radius cutout in the middle. However the wall is drawn, there’s no guarantee that the distribution of paralyzed targets will match the shape of the wall. On the upside, though, it can be sustained to keep dealing damage for as long as targets are immobilized.

A nagpa can’t be confident that a straight wall will pass through more than two out of six targets paralyzed in a 30-foot-radius circle. A solid circle with a 10-foot radius occupies only one-ninth the area of the larger circle, so on average, it’s probably going to cover only one target; a solid 20-foot-radius circle could be predicted to hit three (as we saw looking at fireball), but subtract the doughnut hole, and we’re down to just two.

  • Prismatic spray’s area of effect is a 60-foot cone, which seems great: per the “Targets in Area of Effect” table in chapter 8 of the Dungeon Master’s Guide, both a 30-foot-radius circle and a 60-foot cone can be reasonably expected to include six foes. The trouble is, those two shapes don’t overlap very well. Any given distribution of six targets in a 30-foot-radius circle is unlikely to fall entirely within a 60-foot cone no matter where that cone originates from. It’s possible, but it’s unlikely. Let’s say for the sake of argument that out of six targets paralyzed in a 30-foot-radius circle, the most a nagpa can be sure of is that it can find a point of origin whence a 60-foot cone will contain four of them.

The real question here is which of these spells Paralysis does the most to improve. In other words, which deals more damage, and by how much: fireball against three paralyzed targets, or fireball against four unparalyzed targets? Prismatic spray against four paralyzed targets, or prismatic spray against six unparalyzed targets? Interestingly, wall of fire—if you go by “Targets in Area of Effect”—can only ever be counted on to strike two targets anyway, so casting it against two paralyzed targets seems like pure win. But by how much still matters.

Normally, to keep it simple, I assume a 50 percent chance of making a saving throw against an area-effect spell for the purposes of calculating expected damage. The nagpa’s spell save DC is high, though, so this time I’m going to assume a one-third chance instead.

Cast at 3rd level, fireball deals 8d6 damage on a failed save, for 23 expected damage per target, or 93 total expected damage against four. Against three paralyzed targets, who automatically fail their saves, fireball deals 84 damage. Boosting this spell to 4th level (or any level) increases both those numbers by the same proportion, and therefore doesn’t change the fact that it’s better for regular field use than as a follow-up to Paralysis.

Wall of fire, cast at 4th level, deals 19 expected damage per target, 22 expected damage per paralyzed target. Plus, when you cast wall of fire on a target who can move, they’re usually smart enough to get the heck out before they take more damage—but a paralyzed target can’t, and the soonest they’ll get to make another saving throw to become un-paralyzed is the end of their turn, at which time they’ll get burned again. So really, it’s 19 expected damage per target, 45 expected damage per paralyzed target. But wait—that assumes that the target succeeds on their second save, when the chance of succeeding on the second after failing the first is only 1 in 3. So now it’s 19 expected damage per target, 60 expected damage per paralyzed target! We could go on, but even if we stop here, assuming two targets in the wall, this combo offers a net gain of 82 damage. Astonishingly effective, given how few targets it affects.

Prismatic spray is tricky, because 62.5 percent of the time it deals 10d6 damage on a failed save, 25 percent of the time it deals no damage (but rather imposes a debilitating condition), and 12.5 percent of the time it does damage equal to or greater than 10d6. Trust me when I say this works out to 23 expected damage per target, or 141 expected damage against six targets. Against four paralyzed targets, who all fail their saves, the expected damage is 28 per target, or a total of 112. Once again, contrary to expectations, Paralysis doesn’t improve the result of an immediately following prismatic spray.

What if the nagpa gets a lucky break, and its targets are arranged just right, so that a fireball or prismatic spray could strike all six targets affected by Paralysis? Then fireball deals a total of 168 expected damage, a gain of 75 over its damage against four targets who are free to make Dex saves, while prismatic spray deals 169 total expected damage, a gain of just 28 over its expected damage against six unparalyzed targets. Wall of fire, amazingly, is still the best spell to combine with Paralysis—and that’s without its targets being arranged so fortuitously that it can hurt more than two.

Paralysis recharges only on a roll of 6. It’s entirely likely that the nagpa will get only one chance to use it in a combat encounter. Consequently, it’s not going to waste it: it uses it only when six or more of its foes, or all of them if there are fewer than six, are within 30 feet of it. If it can position itself within 30 feet of this many opponents without putting itself at undue risk, it does so (movement), then uses Paralysis (bonus action) at the start of its turn and immediately follows it up with wall of fire (action), in whatever configuration incinerates as many paralyzed opponents as possible.

Incidentally, this isn’t a bad way to kick off a combat encounter with a nagpa: If you can pull it off, have it use its various charms and deceits to talk all its soon-to-be opponents into standing in a nice, straight line or a perfect circle! “Ah, that’s excellent. Thank you. Now stay right where you are—I mean it!” *fwoom*

To analyze the rest of the nagpa’s spells, I’m going to divide them between those that require concentration and those that don’t. (Except for counterspell, which is cast as a reaction, all of these spells take an action to cast.)

Concentration required

  • Dominate person does just what it says on a failed Wisdom saving throw. Obviously not so good against a cleric or druid, but not so bad against a fighter or rogue who can then be sent to beat up a cleric or druid. However, the target gets a new saving throw every time they take damage, so the control may not last long.
  • Confusion is somewhat weak compared to other things the nagpa can do: its radius is small, generally affecting only two targets, and its results are unpredictable. If the nagpa has many minions, it becomes more effective.
  • Fly makes the nagpa much faster, allows it free movement over difficult terrain, and can lift it out of an otherwise inescapable pit or other trap. Realistically, though, by the time PCs are high enough level to fight a nagpa, they have ways of flying, too, so it’s a poor escape plan and mainly just makes the nagpa a more attractive target for archers. It does lift it out of melee reach, at least, but the opportunity cost (both in casting time and concentration) is high.
  • Hold person is always good, though not necessarily for mopping up enemies that made their saves against Paralysis, because it’s the same save. Boosting it to 4th level to paralyze three enemies is an attractive option.
  • Ray of enfeeblement has the unattractive property of being most useful against those best equipped to resist it. However, the fact that the nagpa’s spell save DC is so high makes it a barely plausible gambit against a melee fighter who’s likely to shake off their paralysis at any moment.
  • Suggestion and detect magic are not good uses of concentration, given what else the nagpa can do.
  • Protection from evil and good is situational: a nagpa will cast it only to defend itself against an aberration, celestial, elemental, fey, fiend or undead of CR 6 or higher summoned by its opponents; or against two of CR 5, three of CR 4, four of CR 3, or six or more of CR 2.
  • Witch bolt might be amusing to boost to 4th level if and only if the nagpa has nothing better to do with its concentration.

Concentration not required

  • Feeblemind takes an enemy out of the fight for good if they fail their saving throw. Since it calls for an Intelligence save, this one’s bad against wizards and great against nearly everyone else. An irksome sorcerer, warlock, cleric or druid casting evocation, transmutation or necromancy spells is the ideal victim (unless they’re a gnome—Gnome Cunning makes them a hard target).
  • Etherealness is the nagpa’s escape hatch. It has to be very careful, because it has only one 7th-level spell slot and one 8th-level spell slot. If it casts both prismatic spray and feeblemind, it has no more gas for its getaway vehicle.
  • Prismatic spray is extremely strong—when the nagpa faces five or six opponents. Against four or fewer, it does no more damage than fireball, at much greater cost. And against seven or more—well, the nagpa may get lucky and hit more than six targets, but it shouldn’t count on it.
  • Circle of death wasn’t included in the Paralysis discussion because it calls for a Constitution save, not a Strength or Dex save. It deals necrotic damage, the same amount as fireball, but over a much larger radius, so that it can potentially affect many more targets—a cool dozen, according to “Targets in Area of Effect.” A nagpa facing a whole horde of foes breaks out circle of death.
  • Disintegrate, on the other hand, is the big gun to pull out against a troublesome front-line fighter or skirmisher or an especially annoying support spellcaster. It deals no damage at all on a successful saving throw but an average of 75 damage on a failure, so the best target is one who’s still struggling to shake off Paralysis or is restrained from being hit by prismatic spray’s indigo ray.
  • Dream and geas both take 1 minute to cast and are therefore part of the nagpa’s scheming kit rather than its scuffling kit.
  • Hallucinatory terrain, which takes 10 minutes to cast, is the same.
  • Counterspell is automatic against a 3rd- or lower-level spell that would require the nagpa to make a Dex or Con save. Against a 4th-level spell that would do the same, the nagpa will upcast it.
  • Charm person is better used outside combat than in it. Corruption has the same effect (although it does last only one round), has three times the range, targets Charisma instead of Wisdom, costs no spell slot, doesn’t require concentration and uses a bonus action.

Paralysis plus wall of fire is the combo the nagpa is hoping to be able to set up, but if its opponents aren’t in the right places to make it work when it’s ready to kick off combat, it sucker-punches either the most threatening-looking enemy caster with feeblemind or (if there are too many threatening-looking enemy casters to choose from) the entire opposing side with prismatic spray or circle of death. If it chooses prismatic spray, it first positions itself where it can catch six or more enemies in the area of effect, or all of them if there are only five (if there are four or fewer, it doesn’t bother with prismatic spray at all). It chooses circle of death only if there are at least a dozen enemy creatures in the area of effect.

After taking this action, the nagpa uses Corruption (bonus action!) to keep one of its foes from bothering it for a round. Since it targets Charisma, bards, paladins, sorcerers, warlocks and snappily dressed rogues are bad bets, as are all elves and gnomes; barbarians, druids, rangers and wizards are pretty good bets; and clerics, fighters and monks fall somewhere in between. (A nagpa can tell the difference between spellcasting types at a glance.) Again, its top priorities are, primarily, to keep its opponents from casting spells that require Dex or Con saves, and secondarily, to keep them from hitting it with sharp or heavy things.

On subsequent rounds, it keeps maneuvering around until it can use Paralysis/wall of fire the way it wants to, taking the Corruption bonus action to temporarily neutralize an opponent whenever it can’t use Paralysis. It doesn’t cast feeblemind if it’s already cast prismatic spray; it doesn’t cast prismatic spray if it’s already cast feeblemind. It saves disintegrate for a paralyzed opponent, but if its foes are taking too long to get into formation for Paralysis like they’re supposed to, it casts hold person at 4th level, against the three targets with the lowest Wisdom (with Intelligence 23, it knows who these are), to be followed up with disintegrate on its next turn. But it doesn’t just cast hold person to make targets easier to hit, or to keep them from doing anything unneighborly. Having two or three enemies frozen in place makes it easier to bait the rest of one’s enemies into position for Paralysis, and that’s just what the nagpa does. If everyone lines up properly while the nagpa is preparing to cast disintegrate, it’s happy to go ahead and cast that spell (action), use Paralysis while the window of opportunity is open (bonus action) and wait until its next turn to cast wall of fire.

The rest is all cleanup. The nagpa keeps wall of fire going for as long as there are at least two foes taking damage from it. When there aren’t, it drops the spell and switches to hold person (if it hasn’t cast that already) or dominate person (if it has). If a criterion for protection from evil and good is met, that spell trumps any other concentration-required spell except wall of fire. It casts counterspell against enemy magic as described above. Even against fewer than a dozen opponents, it doesn’t mind using circle of death after showing off its Paralysis/wall of fire party piece, but first it takes advantage of any lingering paralysis to cast disintegrate (if it hasn’t already) or fireball or to land auto-crits with its staff (only against enemies it doesn’t have to walk through flames to get at).

Nagpas are both shrewd and prudent, and they have no particular desire to risk their necks. They cast etherealness and retreat when only moderately wounded (reduced to 130 hp or fewer). If etherealness is squelched somehow, they try fly (or Paralysis/wall of fire, if they haven’t gotten to yet). If that’s squelched too, they keep fighting until they’re seriously wounded (reduced to 74 hp or fewer), then flee, casting witch bolt (at 4th level if possible, otherwise at 2nd or 1st), fireball, fire bolt (3d10 fire damage!) and, if truly desperate, suggestion to cover their retreat; and continuing to use Corruption to discourage their pursuers.

Next: balhannoths.

15 thoughts on “Nagpa Tactics

  1. I would suggest you give this the same numerical analysis. Replace Fireball with Sickening Radiance, or any other large radius spell. I see no reason why a DM cannot take any monster and modify the spell lists/ attributes of the provided set if they are not optimized. A CR 17 monster is ABSOLUTELY going to have a synergised set of spells that wreaks havoc on players.

    At least one of my DM’s alters the statblocks on monsters, given his table has at least 4 other DM’s who run their own tables, and these guys know monsters.

      1. I did suggest it, yes. You want to get snippy, that is your right. It is your blog. But given the amount of energy you burned writing about the percentage of the creatures caught in a 20 radius versus 30 foot radius, I would think it is trivial.

        And yes, I do believe that a highly intelligent monster recognizes when some spells are far less than optimal, and would replace them with ones that work inside whatever tactics they prefer.

        Oh, and it took all of 7 minutes to do the research and math on Sickening Radiance:

        Barbarians, Fighters, Sorcerer’s all get Prof in Con, so assume they are operating with at least a +11 to their saves. But they are the same chars most likely to fail a save vs Paralysis. Make the assumption then that all 6 of your hypothetical chars get at least one fail in that radiance. That is then expected value of 22 * 6 chars = 132 HP PLUS disadvantage to all ability checks (not to be confused with savings throws). If some fail a second save, movement is halved, which is a huge huge impact.

        And while you were sniping at me about doing more math, you might want to check what you wrote. You either mistyped or miscalculated the damage from Fireballs. Expected value of 8d6 is 28, not 23. And against 4 targets that is 112 damage, not 93. You did get it right for 3 targets = 84.

        1. 7 minutes, eh. Let’s take your little exercise. First, start off by gouging out the part of your brain telling you that what you’re actually asking for is a comprehensive tactical analysis of every spell in the game that a Nagpa can cast vs. every spell in the game that a Nagpa can cast. From the framing of your question and follow up, it seems you are way ahead of us here.

          So 1st- to 8th level wizard spells from only official sources because when I homebrew a monster’s statblock I only look at official content. There are 300 of those.

          Whilst you’ve got your skull cracked open and are self-lobotomising, might as well take out the part of your brain telling you “Surely that 7 minutes time per spell might be compounded by even the tiniest amount when you’re comparing basically every spell in the game for its tactical advantage vs every spell in the game for its tactical advantage” Pop that sucker right out.

          Ok. With your brain surgery complete we can move onto the analysis stage. That’s going to take 35 hours by your calculations. You can do the math yourself. I’ll wait.

          “But I was only saying look at one spell, and it only took me 7 minutes to do the analysis and I even found time to complain about this guy” Good for you. You should go start a blog somewhere with your musings on monster tactics. Sounds like a great idea. Surprised no one has thought of it

          Thanks so much for the blog Keith. It really has been probably one of the biggest helps for my games. The biggest boon it gave me was a frame of reference to think about intelligent enemies and how they would act. It’s helped me massively in my homebrew and reverse engineering what I want out of a monster. The time and effort you put into doing this work is appreciated. And I’ve got your book on pre-order.

          1. Sadly you already said most of what I wanted to say, only more eloquently. I just want to pile one last thing on.

            If you want an optimized wizard enemy (because, the nagpa is basically just a wizard) there are -Scores- of threads on the internet discussing wizard cheese tactics. Tactics that would work even better for NPCs, than PCs, such as simularcum cheese, or animate dead+demiplane combos.

            It sure takes less than 7 minutes to Google that.

  2. The 23 is correctly calculated, based on the assumptions in the post – he says that he’s assuming that targets fail their saving throw 2/3 of the time. 8d6 on average is 28 damage, 8d6 halved on average is 14 damage. (28 x 2/3) + (14 x 1/3) = 23.33333, which rounds to 23.

  3. I’m glad Dinsdale Pirannha has 7 minutes to do the research and math on Sickening Radiance. I have a baby to take care of and two books to get ready for publication. I have to squeeze this in when I can.

    This seems like a good moment to remind everyone that when you comment on this blog, you’re a guest in my house. I like having guests in my house. I like listening to what you have to say. And I don’t mind being corrected. But I can and do place people in a moderation queue if they can’t maintain an appropriate tone, and if they get belligerent about it, I can and do revoke commenting privileges entirely.

  4. > “work in the shadows” and “emerge [only] when they can deliver a finishing blow,”

    As in behind the scenes through proxy agents, not literally skulking through shadows. All you need is a solid social network for that.

  5. I’m gonna defend *Suggestion* in combat.

    The Nagpa as established works from the shadows. This means it has spies and informants. If the party has any disharmony then what a “reasonable suggestion” is widens dramatically.

    “X is planning to betray you, kill them before they can.” can be incredibly effective depending on who the spell target is, and who X is. It’s basically better than *Dominate __* in that regard because it doesn’t end as easily.

  6. It’s worth noting that the Nagpa will never bash with its staff against a Paralyzed opponent for crit damage basically ever. This is because it can blast a Firebolt (or Chill Touch, if its opponents have been healing themselves) from point blank range, and get an autocrit that way without suffering disadvantage on the attack from being within 5 feet of a hostile creature; they’re paralyzed, they can’t impose that disadvantage. An upcast Witch Bolt is even better for this purpose if the Nagpa isn’t concentrating on anything. The only time this doesn’t work is if there’s a non-paralyzed creature right next to the paralyzed one, such that the Nagpa will be in melee with both if it tries this, in which case its better off with the aforementioned Wall of Fire play anyway. The staff whack is literally a last resort attack when the Nagpa is surrounded and out of spells or when it is supremely overconfident and wants to demean its opponents (or servants).

  7. NdukwuIt’s important to know that a spell has to have an attack roll to crit someone, so while Fireball, Disintegrate, Wall of Fire, and Prismatic Spray will hit and they’ll auto-fail, they will not take extra damage. Still good strategies for following up a mass paralysis though. I would also consider having the Nagpa try and dominate someone before the fight. If the Nagpa knows a party is closing in on it, it can use Hallucinatory Terrain to hide itself behind what looks like rubble or something similar and natural. With itself having Truesight it’ll be able to see when the party is closing in and more than likely have the element of surprise unless someone in the party can immediately see through illusions through a magic item, otherwise, no one is going to find some rubble in Shadowfell interesting. That’ll give the Nagpa enough time to pick the best target for Dominate Person, from Fighter, Barbarian, Bard, or Sorcerer. Barbarian will probably be the better choice for the setup since they’ll have a good + to initiative. On the next turn if the Barb goes first, the Negpa will telepathically command it to move more than 30 feet away from it so it is out of the range when it unleashes the Paralysis. Once that happens the Barbarian can start critting its teammates.

    https://twitter.com/mikemearls/status/489204964535713792

  8. This, of course, all assumes that the Nagpa has no choice but to fight. If given any other option, the Nagpa will either not engage or flee at the first opportunity, because they value their skin above all else.

    So, while this analysis is absolutely on point, a DM who roleplays the Nagpa effectively will almost never leave them in a position where they willingly fight at all. A Nagpa who must fight has effectively already lost, from its perspective.

    On the subject of spell choices, while it’s true that a DM can (and should) modify the Nagpa’s “kit” to suit the game, this article is meant to analyse the monster AS WRITTEN. Obviously an encounter with a Nagpa would work very differently given other spells. I would even assume, much like the Skeksis they resemble, each Nagpa has its own specialties and favored spells.

    For that matter, being nigh-immortal, trans-dimensional wizards that regularly pillage the ruins of dead empires, a Nagpa encountered in combat will probably be strapped to the gills with magic items. Up to and including spell scrolls, allowing it to cast one-off additions to its own normal load-out. Given the aforementioned Expert Level Cowardice the Nagpa display, I would personally have at least one extra “escape hatch” option at the ready. For the exact purpose of not finding themselves trapped in a fight.

  9. Just wanted to say thanks for all the work. This analysis and the others are both informative and interesting. 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.