How are the hit points of the Hound summoned with the Hound of Ill Omen feature of the Shadow Sorcerer determined?

I have read over the feature several times now, and I have a question. Here are the features that differ from the MM Dire Wolf stats:

The hound uses the dire wolf statistics (see the Monster Manual or appendix C in the Player’s Handbook), with the following changes:

  1. The hound is size Medium, not Large, and it counts as a monstrosity, not a beast.
  2. It appears with a number of temporary hit points equal to half your sorcerer level.
  3. It can move through other creatures and objects as if they were difficult terrain. The hound takes 5 force damage if it ends its turn inside an object.
  4. At the start of its turn, the hound automatically knows its target’s location. If the target was hidden, it is no longer hidden from the hound.

The question I have is in regard to item 2. It states that the Hound appears with a number of temporary hit points equal to half the sorcery level. Which of the following is accurate?

  1. It only has a hit point total of half the sorcerer hit points.

  2. It has half the sorcerer hit points as temp hit points, on top of the hit points that a Dire Wolf would get per the MM stat block?

Attempt 2: mechanical / balance issues with War Priest feature replacement?

I recently proposed a homebrew replacement for the War Priest domain feature. Consensus was that it was overpowered, because 1st level domain features should (paraphrasing):

  • Be situtational if they are powerful.
  • Have a meaningful resource cost.
  • Pose a trade off to the player.

(My personal opinion is that some equivalent abilities – e.g. the Order Domain feature – do not abide by these restrictions, but obviously there are contextual differences between subclasses).

I’ve written a new homebrew replacement for the War Priest feature based on that feedback. It allows a War Cleric to make an attack as part of a casting action by weaking the spell cast.

When you use your action to cast a spell by expending a spell slot, you may instead have the spell assume the level below the slot’s and target one or more hostile creatures. Make one weapon attack against a creature that was not a target of the spell, as part of the casting action.

The assumed level must be valid for the spell, and the hostile creatures must be valid targets for the spell at that level.

Some comments:

  • It’s situational. It requires the cleric to be both blasting (instead of the more useful healing / buffing) and engaged with other enemies. However, it does remain useful into the mid- and late-game.

  • It’s limited by the number of spell slots the cleric can use. However, it doesn’t have an action economy cost (cf. the 1st level feature of the Death Domain).

  • It trades an attack for one level of a spell slot. So the cleric can either throw away a low-level slot to weak blast + attack, or nerf a high-level slot to strong blast + attack.

  • It stacks with Spiritual Weapon. So in the best case, the cleric can blast + attack + bonus attack with spritual weapon. In contrast, the Order Domain cleric can heal + trigger a sneak attack + bonus attack with spiritual weapon.

Do you think there are mechanical or balance issues to the homebrew feature?

Are there mechanical or balance issues with this War Priest feature rework?

War Priest is a 1st level feature that lets War Domain Clerics use their bonus action to make a second attack when they take the Attack action.

I (and others online) feel that this feature is poor, for a few reasons:

  • The wisdom-based use limit does not scale well to mid- and late-game.

  • The Spiritual Weapon spell available to all Clerics outperforms the feature in most circumstances (e.g. it allows a casting action to be followed with an attack bonus action) with only slightly more resource cost.

  • The feature is hard to use in mid- to late-game, where Clerics almost always cast a spell in preference to using the Attack action.

I would like to propose that my DM remove the original feature and replace it with the following:

If you cast a spell with a spell slot of 1st level or higher and an ally is one of the targets of the spell, you can make one weapon attack as a bonus action.

This replacement is based off the Voice of Authority feature, and seems roughly balanced with it (e.g. it is slightly stronger compared to that feature used on a Wizard, and slightly weaker compared to that feature used on a Fighter).

It also leaves a distinct niche for Spiritual Weapon: when you’d like to all-out blast by using damage spells along with bonus action attacks.

Are there any balance or mechanical issues that you can foresee with this replacement feature?

Comparison of feature importance values in logistic regression and random forest in scikitlearn [closed]

I am trying to rank the features for binary classification, based on their importance using an ensemble method by combining the feature importances estimated by random forest and logistic regression. I know that logisticregression coefficients and random forest feture_importances are different values and Im looking for a method to make them comparable. Here is what I have in mind:

X=features y=lablels rf=RandomForestClassifier() rf.fit(X,y) RFfitIMP=rf.feature_importances_/rf.feature_importances_.sum() #normalizing feature importances to sum up to 1 lr=LogisticRegression() lr.fit(X,y) lrfitIMP=np.absolute(lr.coef_)/np.absolute(lr.coef_).sum() #Taking absolute values and normalizing coefficient values to sum up to 1 ensembleFitIMP = np.mean([featIMPs for featIMPs in [RFfitIMP,lrfitIMP]], axis=0) 

What I think the code does is to take the relative importance from both models, normalize them and returns the importance of features averaged over two models. I was wondering whether it is a correct approach for this purpose or not?

Is the Oath of the Ancients Paladin’s Elder Champion feature intended to still work while unconscious?

The Paladin’s Oath of the Ancients Level 20 Elder Champion feature states:

Using your action, you undergo a transformation. For 1 minute, you gain the following benefits:

  • At the start of each of your turns, you regain 10 hit points.

  • Whenever you cast a paladin spell that has a casting time of 1 action, you can cast it using a bonus action instead.

  • Enemy creatures within 10 feet of you have disadvantage on saving throws against your paladin spells and Channel Divinity options.

Once you use this feature, you can’t use it again until you finish a long rest.

What it does not state is that the transformation ends or you lose the benefits of the transformation when you are unconscious or incapacitated.

If the Paladin is knocked unconscious while it is transformed, does that mean the Paladin will be able to regain consciousness at the start of next turn?

To contrast, the Oath of the Crown’s Exalted Champion includes this statement:

This effect ends early if you are incapacitated or die. Once you use this feature, you can’t use it again until you finish a long rest.

Emphasis mine.

The Paladin’s Auras of Protection specifically state that:

Starting at 6th level, whenever you or a friendly creature within 10 feet of you must make a saving throw, the creature gains a bonus to the saving throw equal to your Charisma modifier (with a minimum bonus of +1). You must be conscious to grant this bonus.

Emphasis mine.

Aura of Courage states:

Starting at 10th level, you and friendly creatures within 10 feet of you can’t be frightened while you are conscious.

Emphasis mine.

This seems to imply most of the Paladin’s Level 20 features continue working even while unconscious. The Oath of Ancients Paladin would be able to regain consciousness at the start of the turn. Similarly, the Oath of Vengeance Paladin would still frighten their enemies despite being unconscious on the ground. The Oath of Devotion Paladin would still give off sunlight and damage enemies despite being unconscious.

Is this Rules as Written and/or Intended?

For the Battle Master fighter’s Know Your Enemy feature, what happens if the target’s AC changes while the target is being observed?

Know Your Enemy (PHB, p. 73-74) is a Battle Master fighter feature that lets you determine whether a creature’s Armor Class is superior, equal to, or inferior to yours.

What happens if a character’s Armor Class changes during the 1 minute you observe them for this feature (such as by donning armor)? Does it take the creature’s Armor Class at the beginning or end of the minute?

Does it consider temporary effects that are added such as shield of faith, mage armor, or a shield being wielded, or just the creature’s base AC?

Is Cross-Domain IP tracking a security feature?

We’re using a cloud based web security (proxy) service which basically takes our web requests, scans them and NATs these requests to their public IP addresses. NAT is being done on a per-session basis, whatever that means in detail.

We now have a problem with a third party service which is essentially doing something like this:

  • User goes to URL A (TLD 1) and logs in there
  • The login then sets a cookie with no domain set (so its only valid for this domain)
  • It 302 redirects the user to another Domain URL B, and sets some token as GET parameter

At this point the session is rejected with an error message that the public IP does not match (for the proxy its a new session, thus a new IP address is used when NATing this to the outside)

For me this looks like a way to authenticate the user on one site and using services on another without implementing some proper authentication framework.

My question:

Is this IP address check adding any (noteworthy) security? This seems to be a setting we cannot change on the security provider side, so we may need to convince the provider to change this model.

Would swapping the Assassin’s Imposter feature for the Scout’s Ambush Master feature be unbalanced?

Inspired by this question: Would allowing the Rogue sub-class (Assassin) to benefit from Supreme Sneak be unbalanced?

The Assassin roguish archetype is considered by some to have very weak class features at levels 9 and 13. The linked question suggests replacing the Infiltration Expert feature with the Thief rogue’s Supreme Sneak feature, which I think fits thematically and mechanically, although a sneaky rogue with proficiency and Expertise in Stealth probably won’t really need it or notice the benefits that often.

In a campaign where the Assassin’s RAW features at levels 9 and 13 won’t really come into play, I like the idea of replacing one of them with Supreme Sneak as the linked question suggests, but that got me thinking: what about the 13th level feature, Imposter?

For reference, the Imposter feature description (PHB, p. 97):

At 13th level, you gain the ability to unerringly mimic another person’s speech, writing, and behavior. You must spend at least three hours studying these three components of the person’s behavior, listening to speech, examining handwriting, and observing mannerisms.

Your ruse is indiscernible to the casual observer. If a wary creature suspects something is amiss, you have advantage on any Charisma (Deception) check you make to avoid detection.

Naturally, the first place I looked to was the 13th level feature of the Thief rogue, but I don’t think Use Magic Device is appropriate because it doesn’t fit with the theme of being an assassin, and it’s also a defining feature of the Thief archetype, hardly the same as a feature about sneaking, which by comparison is quite generic.

So, I considered the other roguish archetypes, looking for something that seemed thematically appropriate, and found the Scout’s 13th level feature, Ambush Master. It sounds like something that fits the flavour of a stealthy assassin, without being too niche like Imposter.

The description of the Ambush Master feature (XGtE, p. 47):

Starting at 13th level, you excel at leading ambushes and acting first in a fight. You have advantage on initiative rolls. In addition, the first creature you hit during the first round of a combat becomes easier for you and others to strike; attack rolls against that target have advantage until the start of your next turn.

This feature seems to synergise well with Assassinate, especially the advantage to initiative rolls… but maybe it synergises with it a bit too well? It’s certainly a lot more combat oriented, and clearly makes the Assassin archetype stronger than RAW, but does it make it too much stronger?

Would swapping the Assassin’s Imposter feature for the Scout’s Ambush Master feature cause any major balance issues for an Assassin?

If I have both advantage and disadvantage, and my target has the Elusive rogue class feature, do I have disadvantage on the attack?

At level 18, Rogues get the Elusive class feature, which says:

No attack roll has advantage against you while you aren’t incapacitated.

However, if you have both advantage and disadvantage, they cancel out, and you are considered to have neither.

That being the case, if you have both advantage and disadvantage, and you attack a level 18 Rogue, does Elusive steal the advantage entirely and leave you with disadvantage? Or do they cancel out before Elusive is applied, causing you to roll a single d20?

How does the Sanctuary spell stack with the College of Glamour bard’s Unbreakable Majesty feature?

The College of Glamour bard’s Unbreakable Majesty feature (XGE, p. 14-15) has a similar effect to the sanctuary spell, although it requires a Charisma save instead of a Wisdom save.

If someone attacks a creature under the effects of both, does the attacker need to make both saves?