Does BetterClaw currently support native hierarchical multi-agent orchestration (not via n8n)? Specifically:
1. Can a master agent autonomously assign tasks, monitor progress, retrieve results and coordinate agents end to end?
2. Can it dynamically create/deploy subagents, beyond preconfigured agents?
3. Can agents directly communicate and share context/memory?
4. Can subagents run in parallel, with the master combining outputs?
5. On Tier 5, do spawned subagents count toward the 30-agent cap? What are concurrency and nesting limits?
6. Can the master run an entire project, review results, request revisions and finish without human input?
7. Which features work today versus being on the roadmap? Please share docs or a demo.

Also, how are credits charged for subagents?