Soft launch feedback: install, first-run, and agent workflow rough edges

[rogerchappel]

CrewCmd is in soft launch. If you tried it, thank you. Sharp feedback is more valuable than praise right now.

Please include:

1. OS, Node, and pnpm versions
2. Install path used: local dev, Docker, Tailscale, external Postgres, etc.
3. Did the README get you running?
4. Where did you get stuck?
5. What did you expect CrewCmd to do?
6. Did personal workspace vs company/shared workspace make sense?
7. Did personal runtime vs shared/company runtime make sense?
8. Did #crew, DMs, threads, and agent tasking behave as expected?
9. Where did you lose trust?
10. What would make you keep using it or contribute?

Known early limitations:

• CrewCmd is early-access software, not enterprise-complete.
• The core is Apache-2.0 open source and self-hostable; a managed service may come later.
• ClawHub browsing/import is still preview-level.
• Governance workflows are foundational, not complete enterprise guardrails.
• Docker/self-hosting paths need more real-world testing.