spyrae/kronos-agent-os ? reverse-engineered prompt

Build me a self hosted AI agent runtime that I can run locally and control from a simple dashboard.

I want it to remember conversations over time, load reusable skills from a workspace, call tools through MCP, and run scheduled jobs like daily summaries or monitoring checks. It should have a command line mode for quick chatting, plus support for Telegram, webhooks, and a local control room so I can see memory, job status, and tool activity in one place.

Please make it conservative by default, with risky capabilities like server operations and dynamic tools turned off unless I explicitly enable them. It should work in a demo mode without needing API keys, but also let me plug in a real LLM later, along with optional memory and swarm style multi agent coordination. If you need current docs for any integrations, look them up online as you go.