SAFETY-47: what a shadow run can show, and the watts on its status row

Not a leak and not a write-path defect. Shadow mode's safety is intact -
the substitution, the re-seed, the authorisation gate and the clamp are
untouched. What was wrong is what three merged documents SAY a shadow run
can show you, and a number on the status row that can be 6.61x what a live
controller would command.

SHADOW MODE CANNOT SURFACE A CLOSED-LOOP DEFECT, AND THE DOCUMENTS NOW SAY
SO. The battery is held at 0 W and the carried command is re-seeded to it,
so every measured input - the meter, the billed-quarter accumulator behind
it, the battery power, the SoC - describes a plant that did not respond. A
closed-loop defect is one whose stimulus IS that response. This is the same
property that makes shadow mode safe, so it is not a gap to close later:
the closed-loop answer comes from the closed-loop rig or from a live run.
SAFETY-10 §5 already conceded this for the grid-following loop. It was
never the loop leg's limitation; it is the general case.

THE "STRATEGIES ARE UNAFFECTED" EXEMPTION WAS TRUE UNTIL SAFETY-45 MERGED,
ONE COMMIT EARLIER. That card sizes the buy leg as min(asked, headroom)
with headroom = (900/remaining) x (ceiling - projected), and `projected`
reads the open quarter's accumulator - the meter - which is a function of
what the battery did earlier in that quarter. Measured on TEST-23 stage 3
at peak_cap_w 1500 / house 1200 / max_w 2000, one hour at 5 s:

  SAFETY-45's limit cycle  NOT visible. 28 onsets live vs 1; grid
                           -455.5..+3200.0 W vs flat at the 1200 W house;
                           guard engaged 410/720 cycles vs 0/720.
  SAFETY-46's SoC churn    NOT visible. The SoC moves 0.000000000 pp.
  TARIFF-14's wrong hour   VISIBLE, and it is the only one: which hour, at
                           what price, clearing what spread - identical on
                           720 of 720 charging cycles at a binding cap.

The exemption narrows to exactly what survives: WHICH hour a strategy picks
and WHY are unaffected; HOW HARD it trades is not. The peak guard's ceiling
and reasons are worth reading from a shadow run; its ENGAGEMENTS are not.

THE NUMBER IS CAVEATED, NOT RE-COMPUTED, AND THAT IS THE DECISION. The
arithmetic is not wrong - 900/remaining is correct for a meter genuinely
under the ceiling, and in shadow the meter genuinely is. It is the right
answer to a question nobody asked. Both surfaces now introduce the watts as
what would have been commanded ON THIS CYCLE, in every state, and carry
shadow.SIZED_FROM_METER whenever a measured billed quarter sized them.
Rejected, and recorded in the decision doc: re-computing it (the recorded
decision would no longer be the decision the shipped path reached, while the
arbiter, clamp, SAFETY-28 gate, log and MQTT all ran on the real one - a
second controller everywhere except the wire); recording nothing (blanks the
row exactly when arbitrage is trading, and the log's edge key is the watts);
and recording the unsized ask beside it (a second number, no warning, and at
the shipped 3500/2000 both figures read 2000 W).

"UPPER BOUND" AND NOT "SMALLER", AND THE DIFFERENCE WAS MEASURED. Charging
only adds offtake, so the shadow size is >= the live size with EQUALITY at
the start of every quarter. Pairwise over 3600 charging cycles in five
configurations: 0 below, 730 exactly equal. The first draft said "smaller",
which would have been false on every one of those 730.

AND THE PREDICATE IS NOT KEYED ON WHETHER THE TRADE WAS CAPPED. arbitrage's
own sentence says "SIZED BY THE CAPACITY-TARIFF GUARD" only when the
headroom came out below the ask - in a shadow run, the START of the quarter,
where the two runs agree to the float. By the end, 900/remaining has
multiplied the headroom past the ask, the trade is not capped, the sentence
is silent, and the record is at its most wrong. A caveat keyed on it would
be quietest exactly where it is needed loudest.

SAFETY-10 §6's COMMITTED EVIDENCE COULD NEVER HAVE FAILED, FOR TWO
INDEPENDENT REASONS, AND CLOSING EITHER ALONE CLOSES NOTHING. Its fixture is
peak_cap_w 15000 against a 350 W house, so the guard never engages and the
sizing is inert - the harness's own comment says so. AND every identity row
reads the FIRST charging sample, where at elapsed=0 the projection IS the
held level and the two runs agree to the float even at a cap that does bind.
Measured: widening the fixture alone leaves that comparison passing. That
second reason is a new shape for MUTATION.md's habit-4 list - an INDEX, not
a fixture - and it is recorded there as the twelfth instance. §6 is re-taken
at a binding cap across the whole run: 716 of 720 charging cycles differ,
headroom 300.0..54000.0 W in shadow against 299.9..307.8 W live, recorded
-2000.0 W against a live -302.4 W.

THE GRADER IS test_arbitrage.py PART 3d, beside SAFETY-45's own closed-loop
regression and built on the same s45_loop() harness, so the runs differ in
one argument. Three lines stand in for _write_unit()'s substitution and
_shadow_settle()'s record and re-seed, marked as a transcription;
test_clamp.py part 8 remains the check on the shipped placement. The section
carries its own precondition into every check: a mutation that stopped
either run trading made it raise on its first min(), kill the file and
report ZERO reds - FCST-06's shape, driven rather than imagined.

AND THE FIX'S OWN FIRST DRAFT CARRIED THE SAME DEFECT. Deleting the caveat
from the action-log line left all 29 files green, because the sentence was
an f-string inside _shadow_settle() and the only suite that drives it uses a
meter_source:off fixture where no arbitrage decision exists. It is
shadow.log_line() now, for the reason words() is a function.

14 mutations, no survivors; M13 and M14 are +0 B and were confirmed on disk
by sha256 with every run purged. M14's first form was aimed at the wrong
direction and correctly survived - min->max makes the shadow record larger,
which is still an upper bound. All 29 check files green.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BwRLw7eKYj2byVdwB6kbjU

goodwe_controller/MUTATION.md

@@ -172,7 +172,7 @@ manufactures them from a **fixture**: the check runs, the assertion is
 evaluated, the sentence above it is true — and the numbers it compares are
 arranged so that the failure it describes could not have changed the answer.
 
-It is not hypothetical and it is not rare. ELEVEN instances in two weeks; four
+It is not hypothetical and it is not rare. TWELVE instances in two weeks; four
 were found by review, two - CFG-04's fixture pair - by an author's own mutation
 battery, which is the cheaper place, one by review again, in the one
 place nobody had thought to point the rule at: not a check in the suite, but a
@@ -414,6 +414,44 @@ same place twice, the check is decoration.
 
   The general habits are **14** and **15** below.
 
+* ⚠️ **SAFETY-47 IS THE TWELFTH, AND IT IS THE SECOND INSIDE A COMMITTED
+  EVIDENCE ARTIFACT.** `DECISIONS/SAFETY-10-shadow-mode.md` §6 tabulated a live
+  run and a shadow run agreeing on `arb_power_w`, on the reason **verbatim**
+  and on the arbiter's winner, and concluded that shadow mode reaches the same
+  decision. **Two independent reasons it could not have gone red, and closing
+  either one alone closes nothing:**
+  * its fixture is `peak_cap_w: 15000` against a 350 W house, so the capacity
+    guard never engages and the thing whose invariance is being claimed —
+    SAFETY-45's sizing of the buy leg from the billed-quarter meter — is in the
+    run and **inert**. ⚠️ **The harness's own comment says so**: *"at this
+    scenario's `peak_cap_w: 15000` it is not binding anyway"*. The rule was
+    written down beside the check it disabled and nobody read it as a
+    disclosure;
+  * and every identity row reads `_sa[0]`/`_la[0]` — **the first charging
+    sample**. At `elapsed = 0` the projection **is** the held level, so the two
+    runs agree **to the float even at a cap that does bind**. Measured:
+    widening the fixture alone — the obvious fix — leaves that comparison
+    passing.
+
+  ⚠️ **THE SECOND REASON IS A NEW SHAPE FOR THIS LIST: AN INDEX, NOT A
+  FIXTURE.** Ask of every identity claim not only *"could this data differ"*
+  but *"**at the point I sampled it**, could it differ yet"*. A comparison
+  taken at `t = 0` between two runs that diverge **because one of them acts**
+  is a comparison of a plant with itself, and no amount of widening the fixture
+  changes that. The replacement, `test_arbitrage.py` part 3d, grades the whole
+  run — 716 of 720 charging cycles differ — and grades the `[0]` agreement **as
+  an expectation**, so nobody rebuilds a check on it believing it covers
+  something.
+
+  ⚠️ **AND THE FIX'S OWN FIRST DRAFT CARRIED A THIRTEENTH.** The new part's
+  action-log check could not reach the log line at all: the sentence was an
+  f-string inside `Controller._shadow_settle()`, and the only suite that drives
+  that method uses a `meter_source: off` fixture where no arbitrage decision —
+  and therefore no caveat — exists. Deleting the caveat from it left **all 29
+  files green**. It is `shadow.log_line()` now. *A sentence assembled at a call
+  site is a sentence no check can grade*, which is CFG-04's substring lesson
+  arriving from the other direction.
+
 Note what these have in common with a masked rung: the mutation *runs*, and the
 suite is *honestly* green. Nothing in a purge, a fingerprint or a disassembly
 can see it, because the code under test really did behave identically on the