First time posting what I've been building.
AI agents shouldn't be stuck in a chat window. They should be on the board with you, working it like a teammate.
So that's what I built. Here it is, working live
@cursor_ai Running agents in parallel is the easy part now. The hard part is checking on them without living in a chat window.
That's exactly what I'm building: a project board agents update natively. Going to try it with this.
@IndieGameJoe That looks so next level, I've also made a tiny 1v1 browser game where you can challenge people to 1v1 your beyblade, 2 days with claude: https://t.co/BeDannGR7K
@enzo_gte I was thinking of just telling it to run opus 4.8 agents but if you have both subscriptions then this can be more efficient if you want to squeeze everything out of both subscriptions
Fable 5 will be hard to use for the most important problems, it refused to audit the security best practices for my own repo and default to opus 4.8
Even crazier it can decide at runtime to silently output bad results, so if I don't get this notification you can assume it may produce nonsense results...
seems like they took a very aggressive reward approach when training 5.5, also there is a small typo "Opus 4.6 grew it.." it should be 4.8
curious to see if they will change the strategy for 5.6, I prefer opus and whenever I see big LOC i just eye roll and want to re-start the session