iCog-Labs-Dev/mettaclaw ? reverse-engineered prompt

Build me a working OmegaClaw agent project that can run as a real chatbot, with the main focus on the Oma Telegram bot experience.

I want it to start cleanly, connect to a chosen LLM provider through environment variables, keep long term memory between runs, and support a continuous agent loop instead of just answering one message at a time. It should be easy to run locally and in Docker, with clear setup steps in the README and sensible default config values. If there are knowledge import steps or preset memory, make those work too.

Please make sure the agent can be extended with plugins or skills, and that the main logic stays simple and understandable. If anything depends on current library behavior or setup details, look up the latest docs online and wire it up so the project actually runs end to end.