jev-pong

# Jev Pong court design Jev Pong simulates a paddle-defense rally: Jev (TypeSafe's System One model) reads the ball's position and velocity every tick and moves the paddle to meet the return. Each hit speeds the ball up, so a long rally outruns the paddle. The agent shapes `pong.json` — the court, the paddle's authority, and the starting ball speed. ## The loop 1. Update `pong.json` (title, courtW/courtH, speed, maxSpeed, accel, and a `style` line that tells Jev a strategy). The viewer watches it and Jev adapts live — no restart, no second server. 2. `node "$JEV_DSH/toolchain/check.mjs"` verifies the workspace's `pong.json` is valid. Run it before you call a court done. 3. Watch the rally. Does Jev return a few balls, speed the ball up, and *sometimes* drop it — or never miss (too easy) / always drop (too hard)? That observation is the finding. ## Reading the court The viewer shows the court, five ghost paddles lit by the probability of each move, a ring where the ball will cross Jev's wall, the paddle's remaining reach, a pace meter with the pace where the paddle is outrun, the text Jev reads, and a bar for every finished rally. Good courts produce a natural arc: Jev holds a few returns, the ball accelerates, Jev scrambles harder, and it finally slips past. `speed` sets the serve pace; `accel` controls how fast each rally runs away; `maxSpeed` is how far a plain paddle move goes per decision (a FAST move goes twice as far). ## Verifying a court `node "$JEV_DS...