diff --git a/.run/P32/t3s3/bank.sh b/.run/P32/t3s3/bank.sh
index 7fcd9fe1e..61e7a262b 100644
--- a/.run/P32/t3s3/bank.sh
+++ b/.run/P32/t3s3/bank.sh
@@ -6,18 +6,19 @@ cd /home/musashi/bfm-decomp
 bin=$1; tu=$2; asmdir=$3; arm=$4; title=$5; shift 5
 fns=("$@"); [ ${#fns[@]} -ge 1 ] || { echo "REFUSED: no function names given after the title (R43)"; exit 2; }
 want=$(cut -d' ' -f1 config/check.$bin.sha); log=.run/P32/t3s3/bank_${bin}_$(date +%H%M%S).log
-for fn in "${fns[@]}"; do d=.run/P32/t3/$arm/$fn.c; [ -f "$d" ] || { echo "NO DRAFT $d"; exit 1; }
+DD=${DRAFT_DIR:-.run/P32/t3/$arm}
+for fn in "${fns[@]}"; do d=$DD/$fn.c; [ -f "$d" ] || { echo "NO DRAFT $d"; exit 1; }
   n=$(grep -cE '\.ent|\.word|__asm__[^;]*"[^"]*\b[a-z]{2,5}[ \t]+\$' "$d" || true); [ "$n" = 0 ] || { echo "VERBATIM-SUSPECT $fn ($n): an asm carrying an instruction (zero-byte fences/launders are fine)"; exit 1; }
   r=$(.venv/bin/python tools/rtu_match.py $fn --split $bin --source $bin --c "$d" --asm-subdir "$asmdir" --work .run/P32/t3s3/verify_bank/$fn 2>&1 | grep -m1 '^MATCH\|^DIFF\|^CC1'); echo "$fn: $r"
   case "$r" in MATCH*) ;; *) echo "STOP: $fn not MATCH"; exit 1;; esac; done
-for fn in "${fns[@]}"; do .venv/bin/python .run/P32/t3s3/splice.py "$tu" "$asmdir" $fn .run/P32/t3/$arm/$fn.c || exit 1; done
+for fn in "${fns[@]}"; do .venv/bin/python .run/P32/t3s3/splice.py "$tu" "$asmdir" $fn "$DD/$fn.c" || exit 1; done
 make build BINARY=$bin -j8 > "$log" 2>&1; rc=$?; echo "BUILD_RC=$rc"
 grep -in 'error\|conflicting\|undefined ref' "$log" | head -6
 [ $rc = 0 ] || { echo "BUILD FAILED — tree left for diagnosis (R42: do not blind-revert)"; exit 1; }
 got=$(sha1sum build/$bin/$bin | cut -d' ' -f1); echo "sha $got vs $want"
 [ "$got" = "$want" ] || { echo "SHA MISMATCH"; exit 1; }
 left=$(grep -c INCLUDE_ASM "$tu" || true)
-body=""; for fn in "${fns[@]}"; do body+="- $fn: draft .run/P32/t3/$arm/$fn.c ($arm agent, agent MATCH re-verified by coordinator rtu_match in the real TU)"$'\n'; done
+body=""; for fn in "${fns[@]}"; do body+="- $fn: draft $DD/$fn.c ($arm agent, agent MATCH re-verified by coordinator rtu_match in the real TU)"$'\n'; done
 git add "$tu" || exit 1
 git commit -q -F - <<EOM || { echo "COMMIT FAILED"; exit 1; }
 $title
diff --git a/docs/backlog.md b/docs/backlog.md
index 23f899644..6592faf94 100644
--- a/docs/backlog.md
+++ b/docs/backlog.md
@@ -2,21 +2,20 @@
 
 > Generated by `tools/backlog.py render` from `.run/backlog.jsonl`. These are functions the Phase-21 automation got **close** on but did NOT byte-match. The whole-binary byte-gate is the sole arbiter (G3/P9): **byte-matches bank and are NOT listed here** — only genuine near-misses/blockers are. Ranked by hand-session priority: **reach** (×N propagation leverage) → **closeness** (match_one mismatch count, lower = closer) → **size**. Each row's `best_draft` is the closest C the machine reached — resume from there.
 
-**Open near-misses:** 14  ·  by status {'near': 13, 'failed': 1}  ·  by class {'WALL-CANDIDATE': 4, 'REGALLOC-PERM': 1, 'WALL-PROVED': 1, 'SCHED': 4, 'REGALLOC': 1, 'ALIAS': 1, 'FRAME': 1, None: 1}
+**Open near-misses:** 13  ·  by status {'near': 12, 'failed': 1}  ·  by class {'WALL-CANDIDATE': 4, 'WALL-PROVED': 1, 'SCHED': 4, 'REGALLOC': 1, 'ALIAS': 1, 'FRAME': 1, None: 1}
 
 | # | addr | reach | class | nins | status | closeness | where it stuck | best draft |
 |--:|------|------:|-------|-----:|--------|----------:|----------------|------------|
 | 1 | func_80032A74 | None | WALL-CANDIDATE | 422 | near | 1 | WALL candidate CONFIRMED in the real TU (S83): 422/422, sole residual idx 244 `lh v0,0x18(s1)` vs target `lhu` — extendhisi2 is a force_not_mem EXPAND (the orphan frame slot is minted only at an lh; §172 producer 3 caller-save area, reload1.c:1445), so lhu loses the 8 frame bytes; ~200 byte-probes + 100-variant retyping sweep (S79) + permuter_ils 8x150s null (S80). Citation current (§172, reload1.c:1445). Draft synced to the TU (typedefs stripped via cdecl.strip_provided_typedefs; D_80064D44/D_8006A970/func_8003F144/func_800316F8 spelled as the TU) | `.run/P32/t4/drafts/func_80032A74_tuclean.c` |
 | 2 | func_80020DA4 | None | WALL-CANDIDATE | 100 | near | 2 | WALL candidate CONFIRMED (S83 rtu DIFF 2): 100/100, phantom 16-byte frame reproduced (address-taken frame_pad[3]); residual = mflo destination $t0 vs $a2 (REGALLOC-PERM), pinning regresses to 79; 5 attempts 51->20->14->8->2 + permuter_ils null (S80) | `.run/S79w/sonnet/func_80020DA4.c` |
 | 3 | func_8017DF28 | None | WALL-CANDIDATE | 119 | near | 2 | WALL candidate CONFIRMED (S83 rtu DIFF 2): expand_block_move copy_addr_to_reg pseudo cse-reused for both later &mtx args (gcc-2.7.2-map cse_expr.md [A23-2]/§H); 119/119; the addiu $s2,sp,0x10 sits in the jal delay slot vs the target bnez slot; five RTL-verified attempts (S71/S79); permuter_ils (S80) "1" was a divergent rewrite (R63). Citation current ([A23-2] present in cse_expr.md) | `.run/S79w/sonnet/func_8017DF28.c` |
-| 4 | func_800CD674 | None | REGALLOC-PERM | 174 | near | 2 | $a3<->$t1 across two masked prim pointers (one shared local can be only one; splitting = +1 pseudo displacing two hoisted constants, 31); SPRT-with-tpage family, §364 mirror (non-struct field stores). Inert: pin order (8 perms), assignment placement (6), volatile index, p+=0x18 spellings, u8* cursor, every pin subset. NEXT: permuter_ils on the pinned seed (§494 recipe) | `.run/P32/t3/opus/func_800CD674.c` |
-| 5 | func_801834A4 | None | WALL-CANDIDATE | 106 | near | 6 | WALL candidate: loop.c movable ordering, closeness 6 (S71); re-probed S83 in the real TU: DIFF 6 for all three stored variants (unchanged) | `.run/S71_gate14/ov_SC03_105-cn-cast-rc-sd/func_801834A4.c` |
-| 6 | func_80011380 | None | WALL-PROVED | 192 | near | 6 | §474 PROVED C-level floor (boot -O0): fold-const.c:882 split_tree merges MULT(MULT(i,2),2); the two escapes each cost one instruction (stupid.c:497 adjacency / expand_decl use-brackets); §388 -O0 colouring oracle. Pinned S79 #8; re-probed S83 in the real TU: DIFF 6 (unchanged) | `.run/m3/opus/func_80011380.c` |
-| 7 | func_800CD92C | None | SCHED | 247 | near | 15 | map §S7 prologue WEAVE: the {sw,lui,ori} groups for 0xE100008D/8F land after the 9-insn li block instead of before — the §17 pins reproduce the ALLOCATION but the hoist happens in sched2. Same SPRT family as func_800CD674 (§364 mirror levers applied) | `.run/P32/t3/opus/func_800CD92C.c` |
-| 8 | func_80039308 | None | REGALLOC | 518 | near | 17 | sched2 + cross-block regalloc: preheader 49/50 swap, un-spellable addu $a2,$a0,$zero (every p=r form cse-propagated), a temp on $t0 vs $s7, and 11 insns of one alias fact (2nd D_80073140[j] load cannot schedule above the D_800C7D20 store from C; /s unlock costs the address allocation, net 20-24). 34->17 via s16 b4 widening copy + dead-local identity sweep (.run/P32/t3/restored/sweep_func_80039308.py) + $2 pin. permuter_ils --klass REGALLOC 2x150s: no gain | `.run/P32/t3/opus/func_80039308.c` |
-| 9 | func_800CF3E8 | None | ALIAS | 469 | near | 27 | ONE cause: the pinned-base alias basin (§500-D1) in the p5/p6 tail; blocks 1-2 byte-exact (idx 0-361). 54->27 via blk2 constant birth order + birthing-boost local w60 + §194-A fence relocation. Inert: all 9 pins load-bearing (+5..+1409), asm position x7, h6 hoist 32x2, tag reshape, P_TAG ADDPRIM, array p6 stores, ~92k annealed variants. Untested: an unpinned alias of p6 for the tag load alone (ONE Opus second look allowed) | `.run/P32/t3/opus/func_800CF3E8.c` |
-| 10 | func_80185810 | None | SCHED | 489 | near | 35 | [permuter] 4 emission windows (see report .run/P32/t3/reports/func_80185810__opus__*.md); exact length, rtu-clean | `.run/P32/t3/opus/func_80185810.c` |
-| 11 | func_8017DC80 | None | FRAME | 346 | near | 46 | the historic -33 LENGTH wall CLOSED (GTE macros must be REAL macros — the TU house block; the splat Handwritten tag is wrong): 346/346, exact 0x70 frame + 9 callee-saved. Residual: reload-slot frame + the la $a0 slot; cse1 unifies OT index and n<4 across func_80010A08(8) (§500-D2 zero-byte asm retire) | `.run/P32/t3/opus/func_8017DC80.c` |
-| 12 | func_800CF408 | None | SCHED | 178 | near | 49 | [permuter] 3 hunks: two prologue sched2 slots, an mlo/mhi allocno tie, a 3-insn block-2 head hoist. Two LENGTH-bearing pins found (tp $17 shared by 0xE1000087/97 = the 6th callee-saved; ob $10 fixes the $t1/$t2/$t3 rotation, 56->49). §351 family (func_8001212C -O0 / func_8017DD04 -O2 exemplars) | `.run/P32/t3/opus/func_800CF408.c` |
-| 13 | func_800CF6D0 | None | SCHED | 249 | near | 137 | sched1 rank_for_schedule last-insn-CLASS tie (every store priority 2, equal refs; QImode stores grouped, loads floated, HImode after — 5 of 6 blocks) + $t1<->$t3 local-alloc swap of the two masks. 249/249 exact length only with tpage-before-len field order (19 swept). Inert at 137: pins on tpage constants/masks, asm re-ties, volatile/memory fences, /s-denial on any store subset, *0x4000 vs <<14, p++ vs p+0x18, / swap. decomp-permuter 122 was semantically wrong (R63) | `.run/P32/t3/opus/func_800CF6D0.c` |
-| 14 | func_80062144 | None |  | None | failed |  | won't compile standalone (loose-typing / missing decl) |  |
+| 4 | func_801834A4 | None | WALL-CANDIDATE | 106 | near | 6 | WALL candidate: loop.c movable ordering, closeness 6 (S71); re-probed S83 in the real TU: DIFF 6 for all three stored variants (unchanged) | `.run/S71_gate14/ov_SC03_105-cn-cast-rc-sd/func_801834A4.c` |
+| 5 | func_80011380 | None | WALL-PROVED | 192 | near | 6 | §474 PROVED C-level floor (boot -O0): fold-const.c:882 split_tree merges MULT(MULT(i,2),2); the two escapes each cost one instruction (stupid.c:497 adjacency / expand_decl use-brackets); §388 -O0 colouring oracle. Pinned S79 #8; re-probed S83 in the real TU: DIFF 6 (unchanged) | `.run/m3/opus/func_80011380.c` |
+| 6 | func_800CD92C | None | SCHED | 247 | near | 15 | map §S7 prologue WEAVE: the {sw,lui,ori} groups for 0xE100008D/8F land after the 9-insn li block instead of before — the §17 pins reproduce the ALLOCATION but the hoist happens in sched2. Same SPRT family as func_800CD674 (§364 mirror levers applied) | `.run/P32/t3/opus/func_800CD92C.c` |
+| 7 | func_80039308 | None | REGALLOC | 518 | near | 17 | sched2 + cross-block regalloc: preheader 49/50 swap, un-spellable addu $a2,$a0,$zero (every p=r form cse-propagated), a temp on $t0 vs $s7, and 11 insns of one alias fact (2nd D_80073140[j] load cannot schedule above the D_800C7D20 store from C; /s unlock costs the address allocation, net 20-24). 34->17 via s16 b4 widening copy + dead-local identity sweep (.run/P32/t3/restored/sweep_func_80039308.py) + $2 pin. permuter_ils --klass REGALLOC 2x150s: no gain | `.run/P32/t3/opus/func_80039308.c` |
+| 8 | func_800CF3E8 | None | ALIAS | 469 | near | 27 | ONE cause: the pinned-base alias basin (§500-D1) in the p5/p6 tail; blocks 1-2 byte-exact (idx 0-361). 54->27 via blk2 constant birth order + birthing-boost local w60 + §194-A fence relocation. Inert: all 9 pins load-bearing (+5..+1409), asm position x7, h6 hoist 32x2, tag reshape, P_TAG ADDPRIM, array p6 stores, ~92k annealed variants. Untested: an unpinned alias of p6 for the tag load alone (ONE Opus second look allowed) | `.run/P32/t3/opus/func_800CF3E8.c` |
+| 9 | func_80185810 | None | SCHED | 489 | near | 35 | [permuter] 4 emission windows (see report .run/P32/t3/reports/func_80185810__opus__*.md); exact length, rtu-clean | `.run/P32/t3/opus/func_80185810.c` |
+| 10 | func_8017DC80 | None | FRAME | 346 | near | 46 | the historic -33 LENGTH wall CLOSED (GTE macros must be REAL macros — the TU house block; the splat Handwritten tag is wrong): 346/346, exact 0x70 frame + 9 callee-saved. Residual: reload-slot frame + the la $a0 slot; cse1 unifies OT index and n<4 across func_80010A08(8) (§500-D2 zero-byte asm retire) | `.run/P32/t3/opus/func_8017DC80.c` |
+| 11 | func_800CF408 | None | SCHED | 178 | near | 49 | [permuter] 3 hunks: two prologue sched2 slots, an mlo/mhi allocno tie, a 3-insn block-2 head hoist. Two LENGTH-bearing pins found (tp $17 shared by 0xE1000087/97 = the 6th callee-saved; ob $10 fixes the $t1/$t2/$t3 rotation, 56->49). §351 family (func_8001212C -O0 / func_8017DD04 -O2 exemplars) | `.run/P32/t3/opus/func_800CF408.c` |
+| 12 | func_800CF6D0 | None | SCHED | 249 | near | 137 | sched1 rank_for_schedule last-insn-CLASS tie (every store priority 2, equal refs; QImode stores grouped, loads floated, HImode after — 5 of 6 blocks) + $t1<->$t3 local-alloc swap of the two masks. 249/249 exact length only with tpage-before-len field order (19 swept). Inert at 137: pins on tpage constants/masks, asm re-ties, volatile/memory fences, /s-denial on any store subset, *0x4000 vs <<14, p++ vs p+0x18, / swap. decomp-permuter 122 was semantically wrong (R63) | `.run/P32/t3/opus/func_800CF6D0.c` |
+| 13 | func_80062144 | None |  | None | failed |  | won't compile standalone (loose-typing / missing decl) |  |
diff --git a/docs/matching-cookbook.md b/docs/matching-cookbook.md
index 5cb251f9f..d2a0f9cd7 100644
--- a/docs/matching-cookbook.md
+++ b/docs/matching-cookbook.md
@@ -37188,3 +37188,29 @@ a2-raw fell to 3 refs / 45 insns (pri 666) and a1-raw (3 / 26, pri 1153) took `$
 an uninitialised temp is R63-unsound — its dead pseudo can be the thing occupying the wanted register; (2) when two parm copies
 swap registers, compute both allocnos' `n_refs`/`live_length` from the `-dl` dump and ask which SOURCE shape adds or removes a
 reference; (3) `docs/gcc-2.7.2-map/regalloc.md` gets the priority formula and the "cross-jump is post-alloc" law.
+
+**§501-C — THE DYING-INPUT SUGGESTION vs THE BIRTHING BOOST: a shared local that dies in two places is not local-alloc's to give,
+and a fresh single-set pseudo is sched1's to glue (P32 T4b, `md_MAIN_009:func_800CD674`, the `$a3↔$t1` REGALLOC-PERM plateau
+banked by a Fable agent — 174/174).** The row: prim 4's masked pointer `and`/`or` wanted `$t1` (= the pinned mask `m24`'s
+register dying at that `and`); one shared `pm2` for prims 3 and 4 gave `$a3` to both; every "split it into two locals" form
+regressed to 31 (S82) and the S83 hand pass measured three short-lived-temp spellings at 31/44. Read in `-dl`/`-dS` + source:
+1. **A pseudo that "dies in 2 places" is excluded from `local_alloc` outright** (local-alloc.c:471 `reg_n_deaths != 1`) → it
+   becomes a GLOBAL allocno, conflicts with hard reg 9, lands in `$a3`. The target's `$t1` is `combine_regs`' hard-reg branch
+   (local-alloc.c:1806–1817): `$9` dying as an INPUT of the insn records `qty_phys_sugg` for the pseudo SET there, tried first
+   in `block_alloc` (:1470–1477) — but it wins only for a pseudo BORN at that `and` (`wipe_dead_reg` clears 9 before
+   `reg_is_set` births the dest). So prim 4 needs its own single-death pseudo.
+2. **Why the split regressed:** `sched.c adjust_priority` (2511–2545) boosts a ready `birthing_insn_p` SET (dest live,
+   `reg_n_sets == 1`) to `max_priority` in the BACKWARD list scheduler — prim 3's fresh `and` gets glued to its `or`, stops
+   filling the load-delay gap after prim 3's second `lhu`, and the `ori $s1,0x97 / lui $s1 / ori $s4,0x96` cascade into the
+   gaps: the old verdict "a fresh pseudo displaces the hoisted constants" was `reg_n_sets` (sched1), not pseudo count (regalloc).
+3. **The zero-byte fix:** prims 3/4 use different variables (`pm3`/`pm2`, single-death each) and `pm3` gets a trailing
+   `__asm__ volatile("" : "=r"(pm3));` after its last store — emits nothing, but its REG_UNUSED second set makes `pm3`
+   `reg_n_sets = 2` (no birthing boost, gap kept) and 2-death (global → `$a3`, as before), while prim 4's `pm2` stays
+   single-set/single-death → `Register 82 in 9`. Measured: fresh pm3+pm2 31; all four fresh 54; inline `(p-0x18)&m24` 175/159;
+   an asm re-setting `pm2` → `$a1` (a volatile asm re-lives all hard regs over the extended qty); asm on `pm3` only → MATCH.
+Map corollaries (regalloc.md / sched.md): `reg_n_deaths` gates local vs global; `reg_n_sets == 1` gates the birthing boost;
+a `__asm__ volatile("" : "=r"(x))` on a variable is a dial for BOTH counters at zero bytes.
+**Process note (coordinator, honest):** two ledger commits (`commit:3962`, `commit:3963`) claimed this bank before it existed —
+the bank helper had been called without the function name (it built the unchanged tree and exited 0) and then with the wrong
+draft directory (it refused). Write the commit message FROM the tool's output, never before it; the helper now refuses an empty
+function list and propagates a failed commit (R43), and takes `DRAFT_DIR`.
diff --git a/phase-ends/CURRENT_PHASE.md b/phase-ends/CURRENT_PHASE.md
index acb487992..20f25175d 100644
--- a/phase-ends/CURRENT_PHASE.md
+++ b/phase-ends/CURRENT_PHASE.md
@@ -99,7 +99,7 @@ Scale estimate: 3–5 sessions.
       `func_80011380` → `.run/m3/opus/func_80011380.c`; `func_801834A4` → `.run/S71_gate14/ov_SC03_105*/`),
       `exclude_audit --write`, `backlog.py render`; a T3 idiom naming a wall's mechanism re-opens that row
       (bounded: one attempt, permuter first). Wall table into this file; cookbook §496+; decision log; accelerators.
-- [ ] **T4b — crack and bank the final 15 (ADDED 2026-09-05 by Drew's directive, supersedes the kill gate: "anything that isn't the original hand-written asm or the PsyQ libs needs to be cracked and banked before we finish"; "before you spawn Fable agents, hand crack each and figure out the blocker to fix it")** (xHigh→Max recommended; Fable agents authorised AFTER the hand pass): for each of the 15 rows (8 NEAR + 7 walls) the coordinator hand-cracks — read the diff + the recorded mechanism, read the compiler (dumps + source), find the lever the record did not try, measure, bank through the verified chain (rtu in the real TU → bank.sh / gate_main) — and only then hands the leftovers to one Fable agent each (brief `.run/P32/t5x/BRIEF.md`). Plumbing landed first: `src/800_c.c` decl/extern edits byte-neutral (this commit). Work dirs `.run/P32/t4b/<fn>/`. **Hand pass DONE 2026-09-05 (S83): all 15 rows read against their diffs and the compiler; 22 spellings measured (all regressions), 0 banks; every row's blocker refined to a named mechanism with the untried lever written in `.run/P32/t4b/<fn>/NOTES.md` (func_80032A74: the transient caller-save area — reload1.c order proven; func_8017DF28: the block-move pseudo reuse vs reorg's branch-target hoist; func_80011380: the target's assignment shape; func_801834A4: sign-hoist vs const-hoist thresholds; …). TU plumbing for func_80039DEC/func_800391D4 committed byte-neutral (`commit:3953`). 15 Fable agents launched ~13:05 MDT (two usage-limit outages, resumed each time via SendMessage). **T4b (1): `main:func_800391D4` BANKED 143dbb89 (`commit:3956`) — a pinned wall: explicit promotion + insn_count pads 7→9 (§501); pin dropped. **T4b (2): `main:func_80039DEC` BANKED 143dbb89 (`commit:3959`) — a pinned wall: the natural switch with duplicated tails, so a2's raw copy keeps 4 refs and wins $a3 (§501-B); pins 7 → 5.** Remaining 13 in flight.**
+- [ ] **T4b — crack and bank the final 15 (ADDED 2026-09-05 by Drew's directive, supersedes the kill gate: "anything that isn't the original hand-written asm or the PsyQ libs needs to be cracked and banked before we finish"; "before you spawn Fable agents, hand crack each and figure out the blocker to fix it")** (xHigh→Max recommended; Fable agents authorised AFTER the hand pass): for each of the 15 rows (8 NEAR + 7 walls) the coordinator hand-cracks — read the diff + the recorded mechanism, read the compiler (dumps + source), find the lever the record did not try, measure, bank through the verified chain (rtu in the real TU → bank.sh / gate_main) — and only then hands the leftovers to one Fable agent each (brief `.run/P32/t5x/BRIEF.md`). Plumbing landed first: `src/800_c.c` decl/extern edits byte-neutral (this commit). Work dirs `.run/P32/t4b/<fn>/`. **Hand pass DONE 2026-09-05 (S83): all 15 rows read against their diffs and the compiler; 22 spellings measured (all regressions), 0 banks; every row's blocker refined to a named mechanism with the untried lever written in `.run/P32/t4b/<fn>/NOTES.md` (func_80032A74: the transient caller-save area — reload1.c order proven; func_8017DF28: the block-move pseudo reuse vs reorg's branch-target hoist; func_80011380: the target's assignment shape; func_801834A4: sign-hoist vs const-hoist thresholds; …). TU plumbing for func_80039DEC/func_800391D4 committed byte-neutral (`commit:3953`). 15 Fable agents launched ~13:05 MDT (two usage-limit outages, resumed each time via SendMessage). **T4b (1): `main:func_800391D4` BANKED 143dbb89 (`commit:3956`) — a pinned wall: explicit promotion + insn_count pads 7→9 (§501); pin dropped. **T4b (2): `main:func_80039DEC` BANKED 143dbb89 (`commit:3959`) — a pinned wall: the natural switch with duplicated tails, so a2's raw copy keeps 4 refs and wins $a3 (§501-B); pins 7 → 5. **T4b (3): `md_MAIN_009:func_800CD674` BANKED d270f695 (`commit:3964`) — the $a3↔$t1 plateau: a 2-death pseudo is excluded from local_alloc; a fresh single-death pseudo for prim 4 + a zero-byte asm set on prim 3's copy against sched1's birthing boost (§501-C).** Remaining 12 in flight.**
 - [ ] **T5 — PhaseEnd** (Max, Tier 1 — prompt R27): P7 checkbox walk; milestone demo (R22 fleet N/N, tools-health,
       `verbatim_check --strict`, final census, wall ledger, parked-5 dispositions via `make audit-disc`,
       `make report` all three metrics + main `143dbb89…` with/without SDK dirs, corrected denominators); **WAIT