Prompt pattern fans out subagents, then loops with a harsh critic until quality passes
paul_cal · x · 2026-07-26
The post says the real substance of the request is tiny: the important part is how to fan out subagents and then verify or critique their outputs.
The attached image shows a prompt pattern for building a high-quality first-person shooter, with instructions to:
- fan out sub-agents for each task,
- loop on each item until it reaches AAA quality,
- use a separate harsh critic to visually compare results,
- keep iterating in Three.js until the result matches or beats the real game.
The main takeaway is a workflow pattern for agent orchestration: parallelize work, add an explicit critic, and iterate aggressively until quality is acceptable.
Related event: Roundup of Multi-Agent Collaborative Coding Workflows(6 posts)→
More from coding & agent
- DeskPilot: open-source native Python desktop client for local LLMs with MCP and sandboxed tools — poofph · 2026-09-23
- Opus 5.5 turns a single image into a Three.js game menu in one simple prompt — majidmanzarpour · 2026-09-23
- Yale PhD student open-sources his paper figure scripts, packaged as a Skill for Claude Code and Cursor — burny_tech · 2026-09-23
- Dev uses Opus 5.5 with Lean to formally verify Claude Agent SDK, yielding 16 bug-fix PRs — bcherny · 2026-09-23
- Chrome flags compromised passwords; user wants a Gemini agent to reset 50+ of them — doooyle · 2026-09-23
- REQL gives coding agents persistent working memory, pipeline views and multi-agent coordination — cl0wnfire · 2026-09-23