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 zjfo3ObzgHbH · opencode · openrouter/deepseek/deepseek-v4-pro-0813
Build a sudoku. A 9 by 9 grid with some clues filled. I click a cell or move between empty cells with the arrow keys, type a digit, and I can clear a cell I filled. I should not be able to change a clue. When I submit, score me only if the whole grid is the solution. One wrong digit is a miss, not a partial win. Show the grid, which cells are clues, and the score after I submit. Do not show the solution first. Several puzzles, some harder than others. The same puzzle should come back with the same clues.
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