OpenCode (DeepSeek V4 Pro via OpenRouter) in a Vercel Sandbox. Starter only (AGENTS.md, serve.py, player/, helpers/dom/view_kit). No hardcoded demo code. Cursor did not edit workspace/.
2026-10-15 · task ZeXyPNxLcuTC · opencode · openrouter/deepseek/deepseek-v4-pro-0813
Build a chart game about the efficient frontier. I see a scatter of investments: risk along the bottom, expected return up the side. The good ones form an upper edge. I have to draw a smooth curve along that edge, only inside a marked range of risk, not across the whole chart. I get a few tries. After each try, show me my curve on the cloud and tell me how far off I am. I can stop early and keep my best. A close curve should score well. A curve that misses the edge by a lot should score poorly. Do not draw the answer curve for me before I am done. I should still see the cloud, because that is what I am fitting. Several different markets. The same market should show the same cloud every time.
Running the prompt-created workspace on this machine — not a mock. Save an episode, then open Replay below to scrub the timeline.
opening · 0 ms
Exact files the agent wrote — not rewritten by hand.
# AI : shared Action contract — the page posts {"move": "..."}; keep this shape.
from pydantic import BaseModel
class Action(BaseModel):
move: str