Show & Tell: Orchestration protocol built with create-meta-prompts

[neoncapy]

Hey! I used your create-meta-prompts skill to build something I think others might find useful.

What I built

A slash command (/orchestrate) that turns Claude Code Opus 4.6 into a zero-read orchestrator with iterative quality control loops. The whole thing was researched, drafted, QC'd, and refined using your meta-prompt pipeline.

The core idea: Claude Code never reads files or reasons about content in the main session. It only plans, delegates to agents, and reads their summaries. Every agent output goes through a QC loop where a QC agent actively hunts for problems (not just validates), then a fix agent applies improvements without dropping anything from the original.

Repo

https://github.com/orangefineblue/claude-code-orchestration-protocol

It's brand new and I just started using it. Would love any feedback on how to make it better.

Why I'm sharing here

Your create-meta-prompts skill was the tool I used to build it. The research phase found patterns from 100+ community repos and official Anthropic docs. Turns out nobody has documented this kind of systematic orchestration protocol before, so I figured it was worth sharing.

Thanks for building these tools. They made the whole process much smoother.