SapienXai/AgentOS — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me an AgentOS style web app that works as a human control panel for managing AI agents and projects from one workspace.

I want a polished dashboard where I can see workspaces, agents, models, runtimes, health status, and recent activity. It should have an onboarding launchpad for setting up OpenClaw or showing clearly when it isn’t available. Let me create or inspect a workspace, build an agent from a guided flow or preset, chat with an agent, dispatch a mission, and review the output or transcript afterward.

The app should feel like an operator console, with a sidebar, main canvas, inspector area, command bar, and planning view. Include screens for adding model providers, creating a workspace from a prompt, and connecting surfaces like Slack, Discord, Telegram, or similar channels.

Use the existing Next.js, React, TypeScript setup in this repo and keep it realistic if live OpenClaw data isn’t available. Show onboarding or fallback states instead of fake “connected” data. Look up current docs online if you need to.

Want more depth? Deep Reverse