I feel like I've summarized it somewhere, but can't find it, so here it is again (it is not finished, I know there are issues left to deal with):

Persons (which includes but may not be limited to paradigmatic adult humans) have rights, which it is wrong to violate. For example, one I'm pretty sure we've got is the right not to be killed. This means that any person who kills another person commits a wrong act, with the following exceptions: 1) a rights-holder may, at eir option, waive any and all rights ey has, so uncoerced suicide or assisted suicide is not wrong; 2) someone who has committed a contextually relevant wrong act, in so doing, forfeits eir contextually relevant rights. I don't yet have a full account of "contextual relevance", but basically what that's there for is to make sure that if somebody is trying to kill me, this might permit me to kill him, but would not grant me license to break into his house and steal his television.

However, even once a right has been waived or forfeited or (via non-personhood) not had in the first place, a secondary principle can kick in to offer some measure of moral protection. I'm calling it "the principle of needless destruction", but I'm probably going to re-name it later because "destruction" isn't quite what I'm trying to capture. Basically, it means you shouldn't go around "destroying" stuff without an adequate reason. Protecting a non-waived, non-forfeited right is always an adequate reason, but apart from that I don't have a full explanation; how good the reason has to be depends on how severe the act it justifies is. ("I was bored" might be an adequate reason to pluck and shred a blade of grass, but not to set a tree on fire, for instance.) This principle has the effect, among others, of ruling out revenge/retribution/punishment for their own sakes, although deterrence and preventing recurrence of wrong acts are still valid reasons to punish or exact revenge/retribution.

In cases where rights conflict, and there's no alternative that doesn't violate at least one, I privilege the null action. (I considered denying ought-implies-can, instead, but decided that committed me to the existence of moral luck and wasn't okay.) "The null action" is the one where you don't do anything. This is because I uphold the doing-allowing distinction very firmly. Letting something happen might be bad, but it is never as bad as doing the same something, and is virtually never as bad as performing even a much more minor (but still bad) act.

I hold agents responsible for their culpable ignorance and anything they should have known not to do, as though they knew they shouldn't have done it. Non-culpable ignorance and its results is exculpatory. Culpability of ignorance is determined by the exercise of epistemic virtues like being attentive to evidence etc. (Epistemologically, I'm an externalist; this is just for ethical purposes.) Ignorance of any kind that prevents something bad from happening is not exculpatory - this is the case of the would-be murderer who doesn't know his gun is unloaded. No out for him. I've been saying "acts", but in point of fact, I hold agents responsible for intentions, not completed acts per se. This lets my morality work even if solipsism is true, or we are brains in vats, or an agent fails to do bad things through sheer incompetence, or what have you.