diff --git a/plans/abstractions.md b/plans/abstractions.md
index bb565e33..9fbba05c 100644
--- a/plans/abstractions.md
+++ b/plans/abstractions.md
@@ -12,6 +12,22 @@ Anything short → _Watching_ (what's missing) or _Rejected_ (why).
 
 ## Last scan
 
+- **Date:** 2026-06-19 (radar loop, pass 40)
+- **Pass 40 — migration critical-path gate has substantially CLOSED (state change).**
+  The migration plan (`loops/migration`, host-readiness.md) treats `lib/host` as the #1
+  gating item — documented "unbuilt, 0/0". That is now stale: the `host` loop (committed
+  44 min ago) has `lib/host` at **10 modules** — `server router handler middleware sxtp
+  page blog feed relations ledger` — passing **181/181**, and is building the *exact*
+  slice-01-blog reference path: `c16924a9` "blog pages as SX trees + render-page (no
+  embedded HTML)", `11bb8c05` "/new mounts the real WYSIWYG editor", `3369166a` per-block
+  guarded render. So the host trio's HTTP/router/handler layer + a blog read **and** write
+  path already exist on SX — well past the "Phase 1 unbuilt" assumption baked into the
+  migration readiness gates. **Operator note for the migration decision:** the
+  host-prereq wave is much further along than the drafted plan reflects; re-baseline
+  host-readiness.md before sizing an implementation loop. (Still open per plan: the
+  multi-`Set-Cookie` host-primitive fix for authed/session slices, and `likes-on-sx`.)
+  Other fleet motion (not abstraction candidates): artdag Phase-7 confluence gate,
+  sx-vm-extensions JIT-in-epoch-serving. Census still EXHAUSTED; no new dedup candidate.
 - **Date:** 2026-06-19 (radar loop, pass 39)
 - **Pass 39 — quiet after an 11-day gap; fleet unchanged.** First scan since pass 38; no
   new candidate. The migration PLANNING loop did useful work *after* its saturation