Directing AI agents by playtesting: one note turned an ugly lighthouse into a beauty
chrisfirst · x · 2026-11-17
chrisfirst describes his "director" role in an agent-built game: play a build, screenshot what's wrong, send notes like "the fence looks stupid" or "can we make the lighthouse look better?", then agents fix it and he replays. The before/after screenshots show the lighthouse transformed from a crude striped block into a proper red-and-white lighthouse with a light beam. Part of his multi-agent game dev project (350k lines of code, 943 agents).
More from coding & agent
- WebArena-Pro: multimodal web-agent benchmark with 20 self-hosted apps, best agent at 62.3% — xhluca · 2026-11-17
- Dev shows 'constellation view' UI where every star is a live agent — kevinkern · 2026-11-17
- The constellation view: monitoring agents as stars — green, yellow, red, gray — kevinkern · 2026-11-17
- Dev directs agents with playtest notes to build a game in four days for Epic MegaJam — chrisfirst · 2026-11-17
- 943 agents, 11.5B tokens, $5,900: 97% of tokens went to re-reading context — chrisfirst · 2026-11-17
- Dev builds stunt-driving game in 4 days with 943 AI agents, 11.5B tokens and $5,900 of compute — chrisfirst · 2026-11-17