README: no-schema-legend in every main-run B/C/D command

README.md

@@ -847,9 +847,14 @@ budget, with the calibrated 512-token reasoning budget:
 
 ```bash
 python -m harness.B.sweep --models all --spatial-code-formats explicit \
-  --input-selections <selection*> --frames <frames*> --reasoning-budget 512
+  --input-selections <selection*> --frames <frames*> --reasoning-budget 512 \
+  --no-schema-legend
 ```
 
+(`--no-schema-legend` on every B/C/D command is the amended main-run prompt design —
+see the prompt amendment in `analysis/preregistration.md`; the legend variants remain
+available as calibration arms only.)
+
 3 configs (3 models × explicit), input selection and frame count fixed from Step 1.
 `format*` = explicit throughout the remaining steps.
 *Hypotheses fed*: H1–H3 (vs A and E); B's extended traces are H23's
@@ -861,7 +866,8 @@ conditioned on H26's certificate.
 
 ```bash
 python -m harness.C.sweep --models all --spatial-code-formats explicit \
-  --input-selections <selection*> --frames <frames*> --reasoning-budget 512
+  --input-selections <selection*> --frames <frames*> --reasoning-budget 512 \
+  --no-schema-legend
 ```
 
 3 configs (one per model) — every non-model axis is now fixed by Steps 1–3.
@@ -870,7 +876,8 @@ python -m harness.C.sweep --models all --spatial-code-formats explicit \
 **Step 5 — Plan D: the ground-truth ceiling, run any time after Step 1.**
 
 ```bash
-python -m harness.D.sweep --models all --spatial-code-formats explicit --reasoning-budget 512
+python -m harness.D.sweep --models all --spatial-code-formats explicit --reasoning-budget 512 \
+  --no-schema-legend
 python -m harness.D.symbolic_eval --spatial-code-format explicit
 ```
 
@@ -896,11 +903,11 @@ python -m harness.A.sweep --models all --frame-selections <selection*> --frames
 # B under the base 16-token protocol (3 runs) -- completes H7's
 # protocol x representation grid
 python -m harness.B.sweep --models all --spatial-code-formats explicit \
-  --input-selections <selection*> --frames <frames*> --base-protocol
+  --input-selections <selection*> --frames <frames*> --base-protocol --no-schema-legend
 
 # C under the base 16-token protocol, winning format (3 runs)
 python -m harness.C.sweep --models all --spatial-code-formats explicit \
-  --input-selections <selection*> --frames <frames*> --base-protocol
+  --input-selections <selection*> --frames <frames*> --base-protocol --no-schema-legend
 
 # The blind floor's extended arm (3 runs) -- the E-base runs happened at Step 1.5;
 # extended-E is the "does reasoning help with nothing but priors" control for H7