A suggestion… Traycer already works beautifully 🙂 To increase both quality and speed of work — some proposals: Agents can cooperate with each other… ok… but often the more eyes that see something, the better. Of course I mean some high‑quality LLMs… For example, you could build an orchestrator and set roles: main model… , next backup … , another model, and there I could plug in, say, Fable High, GPT‑5.6 Sol Ultra, Grok 4.5. And a setting somewhere for whether Traycer should consult proposals, opinions, etc. with the other configured models… and a second setting — when I hit a limit (e.g., Fable) the orchestrator role is automatically taken over by GPT Sol Ultra, etc. If someone cares about speed they can set a fast version because they know Traycer will switch to the next model if needed. Ultimately better quality, fewer errors, protection against rate/usage limits because you don’t have to babysit Traycer. You can do the same for the other workspaces in Traycer so backups are configured… then Traycer could run 24/7.