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 mWHyLnGPjZZS · opencode · openrouter/deepseek/deepseek-v4-pro-0813
Build a formula transcription game. Show me a picture of math. I type the LaTeX that would produce it and submit once. Score me on how close my text is, character by character, with a clear bonus if I match exactly. A huge wrong answer should not be able to score better than a short one that is nearly right, just by being long. Do not show the source next to the picture. Several formulas, from easy to nested. The same picture should come back if I restart that round.
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