diff --git a/docs/re/mission-phase-membership.md b/docs/re/mission-phase-membership.md
index 6076583..bf49e47 100644
--- a/docs/re/mission-phase-membership.md
+++ b/docs/re/mission-phase-membership.md
@@ -206,6 +206,37 @@ The user's correction stands on the data: `UN_e007_ADAN_Turret` is flown by
 30-slot craft formation — never `AI_Structure`. It is an enemy the player fights,
 not a gun bolted to a capital ship.
 
+## 🔴 2026-08-25 — the Turret-biased run was VOID: the guest was frozen
+
+Ran it: `SYLPH_PREFER=Turret`, escort intact (`ast=100.0%`), 1070 s. The phase
+field read **0** throughout and the pilot logged **4029** Turret-targeting
+samples. That looked like a clean negative — "hunting the objective squadrons
+does not advance the phase" — and it is **worthless**.
+
+`frozen.py` says `max_pixel_delta=0`: **the guest was frozen**, and the pilot's
+very first sample at `t=0.0` already shows `spd=0` with the same yaw, pitch,
+target and `d=7186` that it still had at 1070 s. The mission froze on entry to
+flight; nothing was ever shot. `screen_id` said `flight` the whole time, which is
+precisely the trap [mission-freeze-resume-spin](mission-freeze-resume-spin.md)
+documents.
+
+**I was one step from writing this up as a result about the game.** What saved it
+was running the freeze test rather than trusting a plausible-looking log — the
+same lesson that note already carries, arrived at again from the other side.
+
+### ✅ Fixed so it cannot recur silently
+
+`phase_probe.py` now calls `frozen.py` every 60 s and prints
+`*** GUEST FROZEN ***` inline, so a dead-world reading is labelled in the data
+rather than discovered afterwards. ⚠️ `frozen.frozen()` returns a **tuple**
+`(is_frozen, max_delta)` — testing it directly is always truthy and would have
+made the witness fire on every check; it must be unpacked. Verified against the
+frozen guest: the banner fires, and the unpacked value is `(True, 0)`.
+
+**The pilot log is itself a witness** and was not being read as one: identical
+yaw/pitch/target across thousands of samples means a dead world, not a patient
+pilot.
+
 ### 🟡 The prediction to test
 
 If phase 1 clears on those three squadrons dying, then destroying **27 Turrets**
diff --git a/tools/re-capture/phase_probe.py b/tools/re-capture/phase_probe.py
index 3059432..4460de1 100755
--- a/tools/re-capture/phase_probe.py
+++ b/tools/re-capture/phase_probe.py
@@ -15,6 +15,16 @@ This is the first DIRECT observation of a phase advance. Everything about phases
 so far is either static (route names, disassembly) or inferred; nobody has
 watched the number change.
 
+**It also witnesses the freeze**, because without that a frozen run produces a
+perfectly clean-looking negative. Measured 2026-08-25: a run entered flight
+already frozen -- `screen_id` said `flight`, the pilot logged 4029 samples, and
+every one carried the same yaw, pitch, target and distance from t=0.0. The
+phase field read 0 for 1070 s and "no phase advance" was about to be written
+down as a result. It was a fact about a dead world, exactly as frozen.py warns.
+
+So: check `frozen.py` on a cadence and SAY SO in the output. A negative from a
+frozen run is not a negative.
+
 Usage: phase_probe.py [secs] [every_s]
 """
 import sys, time
@@ -51,9 +61,25 @@ def sample():
 if __name__ == '__main__':
     secs = float(sys.argv[1]) if len(sys.argv) > 1 else 600
     every = float(sys.argv[2]) if len(sys.argv) > 2 else 5
+    import frozen
+    next_check = 0.0
     t0 = time.time(); last = object()
     print('singleton ptr 0x%08X, phase at +%d' % (SINGLETON_PTR, PHASE_OFF), flush=True)
     while time.time() - t0 < secs:
+        now = time.time() - t0
+        if now >= next_check:
+            next_check = now + 60
+            try:
+                # frozen() returns (is_frozen, max_delta) -- a TUPLE, so testing
+                # it directly is always truthy and the witness would fire on
+                # every check. Unpack it.
+                dead, _delta = frozen.frozen(5.0)
+            except Exception:
+                dead = False
+            if dead:
+                print('  [%6.1fs] *** GUEST FROZEN -- every reading from here is '
+                      'about a dead world, not a fact about the game ***' % now,
+                      flush=True)
         ph, base = sample()
         if ph != last:
             print('  [%6.1fs] singleton=%s phase=%s' % (