Run a task with AI as a flow of steps you keep, reuse, and refine, not a one-off chat.
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).