It goes negative
This is a case of "the rules don't say otherwise". In particular, the Skills section has nothing to say about results being "low-capped" at 0.
Skills can be further modified by a wide variety of sources—by your race, by a class ability, by equipment, by spell effects or magic items, and so on. [...]
If the result of your skill check is equal to or greater than the difficulty class (or DC) of the task you are attempting to accomplish, you succeed. If it is less than the DC, you fail. Some tasks have varying levels of success and failure depending on how much your check is above or below the required DC. Some skill checks are opposed by the target's skill check. When making an opposed skill check, the attempt is successful if your check result exceeds the result of the target.
I would expect the rule to be found somewhere in this paragraph, but there's nothing.
Do note that some skills have DC 0 examples, like Climb
DC 0: A slope too steep to walk up, or a knotted rope with a wall to brace against.
This DC could never be failed without negative check results, e.g. due to low Strength, or the -5 penalty for accelerated climbing.
The primary source on Favored Enemy is either Player’s Handbook in the ranger class description, or Rules Compendium if you buy its assertion of primacy. The descriptions in the Invisibility description in Dungeon Master’s Guide, Improved Manyshot in Epic Level Handbook, or even darkness, despite also being in the Player’s Handbook, are definitely not the primary source on Favored Enemy.
Further caveats, limitations, and addenda not mentioned in the primary source description are contradictions with that description. If Favored Enemy had defined itself as precision damage, the primary source on precision damage would apply, but it didn’t, which means that neither the precision damage description nor anywhere else can define it as such.
I would be inclined, in general, to follow contradictory rules as far as they go. Darkness and Invisibility cannot define Favored Enemy as precision damage in general, but it can say Favored Enemy doesn’t work in those conditions. Here, specific-trumps-general: rather than trying to redefine what Favored Enemy is (in which primacy asserts itself), they can define a special case which acceptably contradicts the general rules. So my reading of the rules as written would be that Favored Enemy is not precision damage, but it does fail to work in cases of darkness or Invisibility, and does apply only once to Improved Manyshot.
Ultimately, however, I would mostly ignore any and all rules as written that apply any more needless limitations on Favored Enemy. Of all the iconic core class features,1 Favored Enemy is one of the weakest, and that’s even assuming that it “just works” on any and all attacks against the designated foes (including, therefore, Improved Manyshot).
- Slow fall and wild empathy, if counted as “iconic class features,” are definitely weaker than favored enemy, and smite evil and trapfinding give it a run for its money too. That’s still a small list compared to all of the options.
Best Answer
So I went through all the epic skill checks, and categorized them as follows:
Broken—if you can hit the check regularly, you can break the game with the effect. Applies primarily to Diplomacy and Perform (which have the same effects), though Sense Motive is also a problem.
Good—Something genuinely useful at a reasonable DC.
OK—Something maybe useful at a reasonable DC.
DCs—Short for “DCs too high for effect,” applies whenever a check’s effect is just vastly less useful than the amount of investment it takes to make it.
Useless—Effects are useless, even if you can hit the DC.
N/A—Several skills just don’t have any particular epic skill usage.
Some skills are in multiple categories because they have different options and some are one thing and others are another.
In conclusion, there isn’t all that much that’s game-breaking—though it definitely exists—but there isn’t a lot that’s all that good, either. The vast majority of things are just really minor effects at ridiculous DCs.
Anyway, for details on each skill:
Appraise—DCs too high for effect
Very few games even use this skill in my experience—most tables seem to play out the whole shopping and haggling thing just once before deciding it takes way too much time for the entertainment it offers, and skip it thereafter. Even barring that, identify is pretty easily accessible, including in ways that ignore the 100-gp cost (most parties I’m in have at least one of cloistered cleric or artificer’s monocle). So Appraise doesn’t matter and neither does its epic usage. The fact that the DC 50 check gets you only detect magic, not identify, is pretty sad, though.
Autohypnosis—good
All of these are pretty solid, and while the DCs are high, the effects largely justify it. Seems reasonably well-made.
Balance—useless (unless flying is removed from the game, then good)
Balance is a fine skill... until everyone starts flying. Which should be long before anyone is hitting DC 40 checks with regularity.
In a game where everyone agrees to avoid flight—which is my preference—then Balance is fine and the epic extension thereof is fine. But that’s not the kind of game it was published for. An epic usage of Balance really just needed to be a route to achieving proper flight, because otherwise it’s just obviated by flight which everyone absolutely must get by mid levels in order to remain relevant.
Bluff—OK
Bluff is a problematic skill in the ways all social skills are problematic, but the epic extension doesn’t particularly cause any more problems. It’s a little weird that the latter two effects are Bluff and not Disguise, but then the overlap between those two skills has always been problematic. The nonmagical suggestion is fine, possibly even too little for that DC +50 modifier.
Climb—useless
Has all the problems of Balance but much, much worse. Even in a game that chooses to ignore flight, there are so many magical solutions to climbing that are just so much easier than using the skill, that no one should bother with the absurdity of managing a DC 100 check—a 2nd-level spell can manage the same.
Concentration—DCs too high for effect
Irrelevant for the most part, since spellcasters have so many ways out of a grapple and have so much incentive to keep them available. I guess it’s good that it’s possible in a pinch, but I don’t see many characters bothering to push their Concentration this high in order to use it. Honestly, the DC should be a lot less.
Craft—useless
The quick crafting option is part of the basic, non-epic rules for Craft, not a new addition.
The augmented alchemy bit is pointless. I mean, alchemy could really use the improvement, but this isn’t a great solution: it costs way too much for too little benefit, so alchemical items are still useless after low levels; now they just cost you more to make. Then again, unbounded multipliers I suppose could be abused since skills can be pumped pretty easily.
Which, really, is the problems of Epic Level Handbook in a nutshell: the rules are weak to the point of uselessness for ordinary use, but their open-ended nature means that they can heavily reward heavy abuse. It’s kind of the worst of both worlds.
Decipher Script—DCs too high for effect (much too high)
It’s absurd that this effect requires such a high DC. Kind of emblematic of big problems in 3.5e in general: they went to significant lengths to “protect” the value of magic, make sure magical things couldn’t be replicated too easily without magic. There is certainly a limit, but the reality is that it was skills, not magic, that needed protecting. DC 50+ checks to replicate a cantrip is preposterous.
Diplomacy—game-breaking
I mean, this is the big one, obviously, right? We all know this. Diplomacy is a huge mess non-epic, too, but the epic skill checks add “Fanatic” to the list. And the DC isn’t that hard for that effect. Diplomancers break the game into itty-bitty pieces, and the epic skill check is a big part of the reason why.
Disable Device—DCs too high for effect
There are so many easier ways to deal with traps, and those DCs are much too high to really be worth it, but time is sometimes of the essence for adventurers, so I could maybe see the use. Just not enough of it to bother investing sufficiently to make the checks.
Disguise—good
I mean, fine, I guess. Disguise is an interesting skill because while magic can do it better and more easily, in situations where you want to disguise yourself they may very well be checking for magic. Disguise eliminates that concern. And the difficulty of these is mitigated by it being something you only need to roll once, and can prepare for—that makes it so much easier to hit these when you need to. And again, that difficulty turns around and is to your advantage, because people won’t expect it’s even possible.
Escape Artist—DCs too high for effect
Passing through a wall of force is cool, but, ya know, basic teleportation is vastly easier to obtain than the ability to hit DC 120 anything. Particular DC 120 in a nigh-useless skill.
Forgery—good
Forgery is powerful only because it’s opposed by Forgery, which almost no one has. And again, you can prepare it ahead of time, and burn one-off resources to pump the check, so you can manage checks in the 60s or 70s if you really care to, even at low-mid non-epic levels—and even with the reader’s +50 bonus, that should be quite sufficient in most cases.
Gather Information—OK
This is kind of weird, because you would think this would be a fairly straightforward, non-epic application of Bluff and/or Disguise, not a big modifier on Gather Information. I guess it’s nice that you could double-down on Gather Information instead of having to invest in those skills, but those skills are more useful than Gather Information is.
The DC is low enough that I could see using it at low enough levels that divinations don’t completely obviate the skill. Of course, that would most likely be done abusing guidance of the avatar’s +20 bonus...
Handle Animal—DCs too high for effect/OK
Epic Level Handbook is a 3.0e book, and so it missed out on the change in Handle Animal to allow it to be used on Int 1-2 non-animals with a mere +5 DC, rather than the extreme numbers the epic checks suggest for magical beasts. There are magical beasts with Int 3 or more, but Handle Animal should not be used on those creatures at all.
The other option, speeding things up, is interesting, since the gains are massive. Not sure how often that’s likely to come up, but burning some resources to hit DC 50 and train something in a day could be worthwhile.
Heal—useless
Garbage, just utter garbage. Not that that was unique to the epic usage (note the linked question is Pathfinder, and the exceptional cases listed, aside from the 1st-level one, don’t exist in 3.5e).
Hide—good
This is cool. I like this, and the DC seems... probably reasonable enough.
Intimidate—N/A
Gets nothing.
Jump—N/A
Gets nothing.
Knowledge—N/A
Gets nothing.
Listen—DCs too high for effect/good
Defeating illusions is pretty meh; that DC is enormous and it only applies to auditory illusions. The Will save will usually be easier, though you do avoid having to interact.
The special bit about defeating invisibility, though, that’s very useful. And the DC is reasonable! This one is solid.
Move Silently—N/A
Gets nothing.
Open Lock—useless
I mean, knock.
Perform—game-breaking
Same problems as Diplomacy, though Diplomacy is easier to pump to the stratosphere than Perform is. Still, regardless, still really problematic.
Profession—N/A
Gets nothing.
Ride—DCs too high for effect/useless/OK
OK, standing on the mount... why? What benefit does this confer? Nothing is listed, nor is anything coming to mind. Also, that DC is ridiculous: people can do this in real life. People cannot hit DC 40 checks in real life.
The unconscious control thing is meaningless because who is going into combat with a non-combat-trained mount?
Attacking from cover is pretty useful though. DC 60’s kinda nuts, but cover is pretty powerful. Not really worth it but meh.
Search—DCs too high for effect
DC 60 to get only the most minor benefit of detect magic. Terrible.
Sense Motive—DCs too high for effect/game-breaking
These are pretty cool, I guess, but the DCs on the alignment ones are way too high for something replicated by cantrips, even if it is faster. The thought-detecting one is extremely powerful, and not easily replaced, so that’s something that might actually justify a very-high DC. DC 100 though, not sure. Difficult to say—once you can do it regularly, it becomes phenomenally powerful. Falling short, though, is a ton of investment for little. And to be really powerful, you need that bonus all the time, not just once. That’s rough. On the other hand, if you do have it all the time, that could easily break a game. So broken if abused, impossible to use otherwise. Epic rules, everybody!
Sleight of Hand—DCs too high for effect
These don’t seem remotely worth the investment necessary to hit those DCs.
Spellcraft—DCs too high for effect
For the record, the difference between “basic properties” and “all properties” is a 3.0e thing, and not relevant to 3.5e (where identify already does “all properties”). That caused me some confusion.
Anyway, getting around the identify minigame is really easy. Cloistered clerics are extremely common in a lot of parties, and have access to free identify as a 1st-level spell. Artificer’s monocle is a cheap magic item that turns detect magic into a free identify. Both of those things might reasonably not have been anticipated by the authors of Epic Level Handbook, but man these DCs are ridiculously high.
The quick identification thing for alchemical substances is interesting... but most tables don’t enforce the usual rules about alchemy labs and so on for this, which means in my experience most get the “epic” benefit without hitting epic DCs.
Basically, at tables where people are really into the identification minigame, and play those rules strictly, these have some utility, but the DCs are really high. In my experience, there is no need for these effects in the first place.
Spot—good
Better than the Listen versions of the same thing, and hitting DC 40 for an active invisible creature is difficult, but plausible, and worth it. This is one of the best examples of an epic skill check, in my mind.
Reading lips in a language you don’t understand is kinda neat, and the modifier is reasonable. Doing it while moving seems a little unnecessary, but hey, if it comes up, Spot is worth being good at anyway, so this could be a nice bonus.
All-in-all, Spot is probably the best example of epic skill checks.
Survival—DCs too high for effect
Survival is another skill often easily obviated by magic, but at least off the top of my head, I can’t think of any spell that allows the whole party to automatically pass all Fortitude checks against weather. It’s kind of niche, and the DC 60 is more than a bit absurd, but it’s something. If the DC were more reasonable—even just 40—this would be solid.
Why identifying creatures by their tracks is DC 60, though, I don’t think anyone could justify. Without reading Epic Level Handbook, I would have guessed that’s like, DC 20 or so, maybe.
Swim—DCs too high for effect
Cool, I guess, and it certainly seems very hard, but DC 80? I don’t see that really actually getting used.
Tumble—DCs too high for effect/useless
Standing as a free action is cool, though DC 35 seems a bit high for it.
Everything else, though, is useless. Flight obviates both the falling stuff and also the wall climbing, and even barring flight there are still other better options. And since no one should basically ever use total defense—particularly at the high levels where the special rule improving the dodge bonus for doing so becomes available—those rules are useeless too.
Use Magic Device—N/A
Gets nothing.
Use Rope—DCs too high for effect
Does anyone even use this skill? Survey suggests not, because Pathfinder ditched it entirely. Anyway, nothing terribly amazing here, and the DCs are all super-high.