Assassin prereq in core, a few other options outside of core, but Rogue seems the intended entrance. Trades 2 skill point for a good BAB, slightly slows SA. That alone looks fair, now let's take a look at the abilities.

Betrayal: Interesting, but a bit clumsy, many DMs prefer to roleplay NPC interaction more and only use Diplomacy when the PC makes a particular request/suggestion/etc. or on occasion when the conversation doesn't really need to be roleplayed out. Experienced DMs also favor more fluid DCs for Diplomacy checks, and so many DMs don't keep track of NPC state towards the player in those terms, so this would be something else to keep track of. Not reason enough to change it though. Balance wise it seems... okay. Potentially very abusable, but a good DM will keep it in mind.

Neer Failing Light: Part of it necessitates the DM using the "if you no longer qualify you no longer benefit" stand point, which most do. Also not useful to characters who have none of the mentioned feats, but still abusable for players who plan ahead. Overall, fine for the first level, but varying degrees of usefulness, from downright useless, to potentially overpowered.

Willing Possession: Beast is clearly superior because of the DR, even considering how common aligned attacks are. Wording is a bit clumsy. Not much else to say. I don't really like it, but I understand what you're trying to do, and it could work alright. Seems like it'd be a better ability at first level.

Ceremony of Wings: Flight. Not much to say other than: why would they want to dismiss the wings? Doesn't make much sense. There may be some neat cinematic things you could do, but I would just give them flight, or add in a reason why it's a good idea to have them hidden.

Inspire Hatred: This is really great. Best ability so far, well written, clear and easily understood. Only change I might make is to have the effect last longer. I envision this being used by a villain (this is an Evil PRC) to turn a crowd of commoners against a hero or heroes. Assuming the Oathbreaker has... let's see, about a 22 Cha or so, the effect only lasts 6 rounds, which is less than a minute. Hardly one of the great cinematic events you see when such an ability is used. My idea is to make the ability as is usable only twice per day, but add a second piece, usable once per day as a full round action that lasts a number of rounds equal to the Oathbreaker's charisma score. The ones usable twice per day would then be more for distractions while he escapes, while the longer lasting one would be a good combat tactic. I would also add in a clause that says that the ability may not be used to make a target hate itself.

Ceremony of Becoming: Once again, Beast seems superior to me, even for a rogue type, actually especially for a rogue type. Also, "her Con modifier for half" what does this mean?

Pact of Ignorance: Very powerful, odd name. The recharge is a bit inconvenient, a better limitation would just be that it's only usable 1/week or something similar. As is, it draws focus away from the story of the campaign in order to recharge an ability.

Ceremony of Flesh: Once again, Beast is better. The constant DR from Maiden is nice, but non-stacking Natural Armor bonus is not nice. Class features should always stack with the creature's base stats. Regeneration 5 is also superior to DR, especially since the Regeneration is overcome by only Good, and DR by Good and Evil. A possible fix for this would be, instead of giving the two choices a different way to do the same thing, (that is, they both have defensive combat abilities) is to give the two options entirely different strengths. Rework the Maiden choice so that it gives an advantage in another region, say stealth/skills/something along those lines.

Darkness Oer All: Very nice capstone. Good for dealing with spellcasters, and not useless against meleers. I might add that spells targeting the area within the aura have the failure chance even if the caster isn't within the aura.

The class overall is... okay. Few of the abilities need to be moved, it's doesn't have any really apparent balance problem, but it could still use some work as noted above. I'd say... 5 or 6 out of 10. Again, tentatively, I'm really bad at estimating those kinds of numbers.