cubetribe/openclaw-godmode-skill ? reverse-engineered prompt

Build me an OpenClaw skill for Claude Code that acts like a multi agent coding coach.

I want to be able to say a normal request like “add login with JWT” or “fix the checkout bug” and have the skill automatically route the work through the right specialist agents, like research, architecture, implementation, review, testing, documentation, and release help. It should feel autonomous, where I describe what I want and the system figures out the steps.

Make it fit OpenClaw’s skill system and keep it documentation first, since this is meant to be loaded by Claude Code. Include clear setup instructions, the different workflows for new features, bug fixes, API changes, research, and releases, plus a simple guide for how the agents are supposed to work together. If you need to check the latest Claude Code or OpenClaw docs online, go ahead and do that.