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 vsZUa0wImWG3 · opencode · openrouter/deepseek/deepseek-v4-pro-0813
Build a desktop task where I have to leave a short log on the desktop: a file with two specific lines, in order, each ended properly. One line, or the lines swapped, or an extra blank line, should not pass. I can use a terminal or an editor. The score looks at the file when I say I am done. Show a real desktop screen and record it for playback. Restarting starts clean, with the same two lines still to write.
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