Dori switches to OmO as agent engine

This is a meaningful architecture split: Dori handles conversational control while OmO handles multi-agent orchestration. The model is compelling, but reliability at session handoffs and merge gates will determine whether it feels like a super agent or merely a clever wrapper.

• –OmO supplies the coordination layer for parallel agent sessions and model routing.
• –Dori isolates each job in its own session, reducing context collisions across concurrent coding tasks.
• –Messenger-first control through Slack, Discord, and Telegram makes agent work accessible away from the terminal.
• –Evidence-gated completion and merge checks are stronger than trusting an agent’s “done” message.
• –The project remains experimental and early-access, so operational reliability is still unproven.