What should a self-hosted agent runtime guarantee?

[Weegy]

If you self-host infrastructure, what are the non-negotiable guarantees an agent runtime must give you before you'd trust it in production?

To start the conversation, here's what omadia commits to today — tell me what's missing or wrong:

• Data ownership: single-tenant, bring your own LLM key, all data stays on your infrastructure (containers or plain Node stack).
• Isolation: each agent/orchestrator has its own memory namespace + knowledge-graph slice.
• Data scoping: tool outputs are treated as an untrusted boundary (privacy guard).

What would you add — auditability? deterministic replay? offline/air-gapped operation? per-tool egress controls?

⭐ Star the repo if you want to follow the public preview.