Is this completely separate product from Agent Zero? Is the intent to have Space Agent as a (frontend) browser resident agent that can communicate with an Agent Zero (back end) instance? If I were building an Intercom-style, script-injectable widget that operates as a frontend extension of Agent Zero using Agent Zero's internal API, do I now have a more powerful alternative to that architecture? This Space Agent is causing my brain to experience a cognitive over pressure interrupt vector in primary and secondary reasoning chambers... Anyone have answers to these questions?!?!....

What username/password to use for Space Agent? Or, should we "Use Guest Account" for now?

@Marco Chillon Hi Marco, apologies for the delayed reply. I admit that I don't have a deep grasp of your particular issue but I encountered a situation where following an A0 upgrade, my custom model configs reverted to default OpenRouter configs. Agent Zero has a built-in mechanism to override default settings via environment variables with the A0_SET_ prefix. These are checked before hardcoded defaults. A0_SET_ variables provide initial default values when settings.json doesn't exist or when new settings are added. Once a value is saved in settings.json, it takes precedence. They act as a safety net for defaults — catching any settings that get reset during an upgrade. So my solution was to use A0_SET_ variables (added to .env) to override default config reversion and ensure custom configs persist container stops/restarts, example below: # ══════════════════════════════════════════════ # PINNED MODEL CONFIGURATIONS — survive upgrades # These A0_SET_ vars override hardcoded defaults in settings.py # ══════════════════════════════════════════════ # Chat (Main) Model A0_SET_chat_model_provider=anthropic A0_SET_chat_model_name=claude-sonnet-4-5-20250514 A0_SET_chat_model_ctx_length=200000 A0_SET_chat_model_ctx_history=0.57 A0_SET_chat_model_vision=true A0_SET_chat_model_kwargs={"temperature": 0.2, "max_tokens": 8192} ══════════════════════════════════════════════ This approach can be applied to model settings as well as agent profile settings. Anyway, hope this helps you in some way...let me know if it scratches your itch. Dale

If anyone is struggling with migrating from a dual A0 (host/container) setup to a single container + CLI Connector setup, then I feel your pain. Just completed that process in the middle of an ongoing project and I have to say that it was well worth the effort. Prior to the v1.9 release I had synchronized the host/container codebases so that made the migration significantly easier. Anyway, if anyone is attempting this migration and gets stuck just ping me, maybe some aspects of my experience can be helpful to you.

This is interesting....any updates on it?