How do I prevent state bleed in a Unity AI state machine?

[MatthewDG]

I’ve been trying to implement a proper state machine for my enemy AI in Unity, but I keep running into issues where states seem to bleed into each other, like the enemy getting stuck in an attack animation when it should have transitioned to chase. My transition conditions are just a mess of boolean checks and it’s becoming a nightmare to debug or add new behaviors.

[Victoria.M]

Yeah, this happened to me last month. The attack state kept looping because the transition condition stayed true even after the attack started, so the AI never left the animation. I added a short exit delay and a guard that disables transitions while the attack anim is playing. It helped, but it still felt brittle if the anim length changed.

[LunaM]

I tried a more rigid setup once, with idle, chase, and attack as distinct states and a tiny stateIndex to keep them separate. It helped a bit, but the booleans still tangled whenever I added a stun or retreat.

[Aaron49]

I started by blaming the state machine, but then drifted into the movement code and navmesh. Sometimes the agent would hit a waypoint and the anim would still be playing, so the machine looked stuck even though the logic was fine. I ended up adding a small check to finish the current animation before letting the next state switch, and I still keep an eye on root motion.

[Stella.H]

Question: could the real issue be the animation events firing at the wrong moments? I once saw an attack event fire while we were supposed to be chasing, and that made the whole state chain feel broken.