Roadmap
The practical direction for AverQel after the current DeepSpace hardening work: keep the native architecture strong, improve maintainability, and extend safely.
Now: Stable Native Runtime
Preserve the current AverQel-native DeepSpace, orchestration, memory, provider, and connector ownership instead of replacing it with an external SDK runtime.
Next: MCP Standardization
Keep moving connectors toward a more MCP-centric model where dynamic tool discovery reduces hardcoded per-service maintenance.
Next: Operator Acceptance
Continue browser-level operational audits for research, coding, approval, and proactive mission types so documentation and runtime stay aligned.
Later: Team And Platform Layers
Shared provider control, richer workspace ownership, and deeper fleet tooling should come only when they solve real product needs.
Not Current Scope
A risky rewrite of the current runtime is not the goal. The path is additive hardening, better docs, safer observability, stronger runtime contracts, and maintainable connector evolution.
The immediate focus remains stability, streaming clarity, connector control, local/cloud provider flexibility, and agent execution parity across the main DeepSpace workflow.