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 OK07qbfYtvED · opencode · openrouter/deepseek/deepseek-v4-pro-0813
Build an online shop I can click through. Each round asks for a kind of product and a price cap. I can search, open a product, read reviews, add it to the cart, and check out. When I am done, score the order: did I buy something that matches, under the cap? The particular pages I visited should not matter. Buying the wrong thing, or spending too much, should fail. Show the shop pages and the score. Clicking is how I play. Different rounds ask for different products. The same round should start with the same catalogue and an empty cart.
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