homelab client add/remove: surgical inventory edits (preserve comments)

yaml.safe_load + safe_dump stripped every comment from inventory.yaml
on each enrollment, eroding the file's documentation value. New
_inventory_set_age_pubkey / _inventory_remove_host / _inventory_append_host
helpers do line-based edits so comments outside the modified region
survive. inventory.yaml gets its top-of-file conventions block back.

build_host_files.py round-trips cleanly (--check returns 0).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

bin/homelab

@@ -126,6 +126,117 @@ SHARED_SECRETS = [
 ]
 
 
+# -------- comment-preserving inventory.yaml edits --------
+# yaml.safe_load + safe_dump round-trips strip every comment, which is fine
+# for hosts/*.yaml (generated anyway) but unfriendly for inventory.yaml where
+# we want the doc comments at the top + per-section to survive. These helpers
+# do line-based surgical edits instead.
+
+import re as _re
+
+
+def _find_host_block(lines: list[str], name: str) -> tuple[int, int] | None:
+    """Return (start, end_exclusive) line range for `  <name>:` in inventory.yaml.
+    Range covers the host's body up to the next 2-space-indented key (next
+    host) or the next 0-indented top-level key, whichever comes first.
+    """
+    target_re = _re.compile(rf"^  {_re.escape(name)}:\s*$")
+    host_at_2sp = _re.compile(r"^  [A-Za-z][A-Za-z0-9_-]*:\s*$")
+    toplevel = _re.compile(r"^[A-Za-z]")
+    start = None
+    for i, line in enumerate(lines):
+        if start is None:
+            if target_re.match(line):
+                start = i
+            continue
+        # We're inside the target block; look for the next sibling host
+        # or a top-level key to mark the end.
+        if host_at_2sp.match(line):
+            return (start, i)
+        if toplevel.match(line):
+            return (start, i)
+    if start is not None:
+        return (start, len(lines))
+    return None
+
+
+def _inventory_set_age_pubkey(name: str, pubkey: str) -> bool:
+    """Surgically set hosts.<name>.age_pubkey in inventory.yaml. Preserves
+    every comment outside the modified line. Returns True if updated, False
+    if the host block or age_pubkey line wasn't found.
+    """
+    lines = INVENTORY.read_text().splitlines(keepends=True)
+    block = _find_host_block(lines, name)
+    if block is None:
+        return False
+    start, end = block
+    pubkey_re = _re.compile(r"^(\s+age_pubkey:\s*).*$")
+    for i in range(start, end):
+        m = pubkey_re.match(lines[i])
+        if m:
+            lines[i] = f"{m.group(1)}{pubkey}\n"
+            INVENTORY.write_text("".join(lines))
+            return True
+    # No existing age_pubkey line; insert one at the end of the block.
+    # Use 4-space indent (matching the other host fields).
+    insert_at = end
+    while insert_at > start and lines[insert_at - 1].strip() == "":
+        insert_at -= 1
+    lines.insert(insert_at, f"    age_pubkey: {pubkey}\n")
+    INVENTORY.write_text("".join(lines))
+    return True
+
+
+def _inventory_remove_host(name: str) -> bool:
+    """Delete the hosts.<name>: block. Preserves comments outside the block."""
+    lines = INVENTORY.read_text().splitlines(keepends=True)
+    block = _find_host_block(lines, name)
+    if block is None:
+        return False
+    start, end = block
+    del lines[start:end]
+    INVENTORY.write_text("".join(lines))
+    return True
+
+
+def _inventory_append_host(name: str, kind: str, os_name: str, role: str,
+                           netbird_fqdn: str) -> bool:
+    """Append a new hosts.<name>: block at the end of the hosts: section.
+    Returns True on success, False if no hosts: section was found.
+    """
+    lines = INVENTORY.read_text().splitlines(keepends=True)
+    # Find the line "hosts:" at column 0.
+    hosts_idx = None
+    for i, line in enumerate(lines):
+        if line.startswith("hosts:"):
+            hosts_idx = i
+            break
+    if hosts_idx is None:
+        return False
+    # Find the end of the hosts: section (next top-level key or EOF).
+    insert_at = len(lines)
+    for i in range(hosts_idx + 1, len(lines)):
+        if _re.match(r"^[A-Za-z]", lines[i]):
+            insert_at = i
+            break
+    # Trim trailing blanks before insertion point so we don't double-space.
+    while insert_at > hosts_idx + 1 and lines[insert_at - 1].strip() == "":
+        insert_at -= 1
+    block = (
+        f"  {name}:\n"
+        f"    kind: {kind}\n"
+        f"    os: {os_name}\n"
+        f"    role: {role}\n"
+        f"    mesh:\n"
+        f"      netbird:\n"
+        f"        fqdn: {netbird_fqdn}\n"
+        f"    age_pubkey: \"\"\n"
+    )
+    lines.insert(insert_at, block)
+    INVENTORY.write_text("".join(lines))
+    return True
+
+
 def _add_recipient_to_sops_policy(sops_path: Path, path_regex_pattern: str, pubkey: str) -> bool:
     """Append `pubkey` to the `age:` list of the .sops.yaml rule whose
     `path_regex:` line contains `path_regex_pattern`. Preserves comments.
@@ -527,15 +638,8 @@ def cmd_client_add(args: argparse.Namespace) -> int:
         netbird_fqdn = input(f"  netbird FQDN (default: {name}.netbird.selfhosted): ").strip() \
             or f"{name}.netbird.selfhosted"
         role = input("  role (e.g. primary-dev, dev): ").strip() or "dev"
-        entry = {
-            "kind": kind,
-            "os": os_name,
-            "role": role,
-            "mesh": {"netbird": {"fqdn": netbird_fqdn}},
-            "age_pubkey": "",
-        }
-        inv["hosts"][name] = entry
-        INVENTORY.write_text(yaml.safe_dump(inv, sort_keys=False))
+        if not _inventory_append_host(name, kind, os_name, role, netbird_fqdn):
+            die("could not locate 'hosts:' section in inventory.yaml")
         push_inventory(f"client-add: {name}")
         print()
         print("Next steps:")
@@ -549,8 +653,8 @@ def cmd_client_add(args: argparse.Namespace) -> int:
     if name not in inv["hosts"]:
         die(f"{name} not in inventory — run 'homelab client add {name}' first (no --finalize-pubkey)")
     pubkey = args.finalize_pubkey
-    inv["hosts"][name]["age_pubkey"] = pubkey
-    INVENTORY.write_text(yaml.safe_dump(inv, sort_keys=False))
+    if not _inventory_set_age_pubkey(name, pubkey):
+        die(f"could not find hosts.{name} block to update age_pubkey")
     print(f"set age_pubkey for {name}")
     print("granting shared secrets...")
     _grant_shared_secrets(pubkey)
@@ -584,9 +688,9 @@ def cmd_client_remove(args: argparse.Namespace) -> int:
         if not confirm(f"proceed removing {name}?"):
             return 1
 
-    # 1. Inventory
-    del inv["hosts"][name]
-    INVENTORY.write_text(yaml.safe_dump(inv, sort_keys=False))
+    # 1. Inventory — surgical block delete (preserves comments)
+    if not _inventory_remove_host(name):
+        die(f"could not locate hosts.{name} block to delete")
 
     # 2. SOPS — remove the pubkey from shared-secret rules and re-key.
     if pubkey:

inventory.yaml

@@ -1,3 +1,24 @@
+# Homelab inventory — canonical structured topology.
+#
+# Single source of truth. hosts/*.yaml is generated from this file by
+# mcp/build_host_files.py; do NOT edit those by hand.
+#
+# Conventions:
+#   - hostname keys MUST match the actual `hostname` of the machine (on
+#     macOS: `scutil --get LocalHostName` if set).
+#   - `os:` one of: linux, macos
+#   - `kind:` one of: proxmox-host, lxc, vm, workstation
+#   - `mesh:` lists addresses the host is reachable at. Both `netbird` and
+#     `tailscale` are accepted during the migration (see infrastructure/mesh.md).
+#     Prefer netbird FQDNs over raw IPs.
+#   - `age_pubkey:` provisioned by secrets-issuance on first bootstrap and
+#     committed back via `homelab client add --finalize-pubkey <key>`.
+#   - When a service moves hosts, update only the `services:` section here;
+#     never duplicate addresses elsewhere.
+#
+# `homelab client add/remove` does surgical line-edits — comments survive.
+# Avoid round-tripping the file through yaml.safe_dump (it strips comments).
+
 mesh:
   primary: netbird
   accepted: