hostwarden-heinzel-takeoverlisted

# hostwarden-heinzel-takeover Takes hosts over from Heinzel so that each ends where a first onboarding by Hostwarden would have left it. Two phases: 1. **Copy** (steps 1–8) — this clone and the old checkout, nothing else. No server is contacted. 2. **Onboard** (steps 9–10) — each host taken over in turn, through `hostwarden-onboard` as its first Hostwarden connection: read-only, one `read-only:` journal line per host. This is the default; the user does not have to ask for it. **Only in an operations checkout** (`AGENTS.md` → Development or Operations). Access lists and machine memory copied from Heinzel are operations state, and a development checkout can neither keep them nor onboard a host: there, say so, name the operations checkout — the main checkout of a worktree, or one `bin/hostwarden-init` sets up — and stop. `bin/hostwarden-heinzel-takeover` refuses to copy there as well. **Only on an explicit request** — the user naming their old checkout, or `/hostwarden-heinzel-takeover <path>` in Claude Code. A session that merely mentions Heinzel is not a request, and neither is a question about what a takeover would do: answer it, don't start. What makes this safe is not the trigger but the gates: the old checkout is only read, anything this clone already holds is reported before it would be overwritten, the copy contacts no server, onboarding only reads, and the one write into the old tree (step 7) needs its own yes. ## Why an inventory instead of a migrati