The most advanced agent coding setup I've used so far:
- LLM: deepseek-4.1-flash (via @ollama cloud)
- Harness: @opencode
- UI: openchamber
- Skills: superpowers, wayfinder
Crushes through the backlog like an army of ants
All you #indiedev folks need to know about what players want:
Counter Strike 2 ~ 550k players online
Dota 2 ~ 410k online
Egg stealing game on Roblox ~ 1.5 million online
You probably miss an opportunity of your life trying to impress noble Steam players.
I now understand why games stop being supported.
At some point you grow professionally and you see flaws in the game you've made some time ago - and you want to fix the flaws. But...
How do you manage to do more than 2-3 tasks with AI agents at once? It takes me 30 minutes to spec a task an agent will implement twice faster. I see no way to do dozens of tasks at once with meaningful outcome.
