@nexus-cortex/cli 4.124.56 → 4.124.57

package/.env.defaults

@@ -1157,7 +1157,15 @@ CORTEX_ENDTURN_GATE=true
 # once measured. Why not TURN_STATUS: measured neutral-to-positive on short tasks (cwbts 22 vs cwbmv 25).
 # [LEDGER 2026-10-05 @4.124.56] DARK — BUILT + unit-tested (endTurnResolver.test.ts R233 block: floor off/on, no deadline, cap/evidence
 #   collapse). NO bench arm yet; first arm = MiMo bundle + CORTEX_BUDGET_FLOOR_S=1200 (expect ≈ cwbmv 25/40).
+# [LEDGER 2026-10-04 @4.124.56 first arm] FIRED 68/68 (cap 2, evidence cap 1) and OVER-CORRECTED on MiMo-40: 20/40 vs bundle 17 and vs
+#   bundle−MAX_VETOES 25; accept-with-gap doubled (27 vs 13), judge FA 36%. Collapsing the R160 cap + R173 rules is the harm; v2 = clamp the
+#   evidence cap only (CORTEX_BUDGET_FLOOR_MODE=evidence, default). Ledger r-mimo64-bundle-2026-10-04.md §7.
 #CORTEX_BUDGET_FLOOR_S=
+# CORTEX_BUDGET_FLOOR_MODE — R233 v2 (2026-10-04). evidence (default): under the floor ONLY the R166 evidence cap collapses to 1; the R160
+# budgeted cap and the R173 gap rules keep their fraction logic. all: the v1 behaviour (every fraction rule sees below-budget) — measured
+# 10-04 to over-correct (cwbfl 20/40 vs bundle−MAX_VETOES 25/40: accept-with-gap doubled). Events carry budgetFloorMode.
+# [LEDGER 2026-10-04 @4.124.57] DARK — BUILT + unit-tested; first arm cwbfl2 = MiMo bundle + FLOOR_S=1200 (mode evidence), expect ≈ 25/40.
+#CORTEX_BUDGET_FLOOR_MODE=
 # (no lever) R161 HB-JUDGE-TASKSHAPE (4.112.0, 2026-09-16): `isTaskShaped()` gated the EndTurn finish judge and the deadline-exit mentor
 # with a fixed coding-verb list — prompts like "Eliminate cumulative layout shift…", "To determine the beta activity…", "You are given …
 # LC-MS/MS data…" were NOT task-shaped, so 3 of the 7 patch-cell sessions (one a 353-turn finish that shipped broken) were never
@@ -1223,6 +1231,8 @@ CORTEX_ENDTURN_GATE=true
 # [LEDGER 2026-10-01 @4.124.40] DARK — BUILT + unit-tested (endTurnResolver.test.ts R208 cases). NO bench arm yet.
 # [LEDGER 2026-10-04 @4.124.55 cell] FIRST REAL ARM (R208–R214 together, on the bundled MiMo base cwbr vs cwb): 20 vs 17 (+3, flips both
 #   ways → undetermined); engaged (task_rule_plan 40, task_rule_judge 74). R208/R209 inert on MiMo by construction. Ledger r-mimo64-bundle-2026-10-04.md.
+# [LEDGER 2026-10-04 TB4 arm] t4r (c23 w stack + R208–R214, 14 c23 tasks, halted by the key-wide budget cap at 11/14): 3/11 vs c23 g 6/11 /
+#   w 5/11 on the same tasks — no lift, negative-leaning; second-attempt chain opened on 1 task. Ledger r-tb4-valA-2026-10-04.md. Stays dark.
 #CORTEX_VERIFY_SCALE=
 # CORTEX_RULE_INPUTS / CORTEX_RULE_INTERP / CORTEX_RULE_HOLDOUT / CORTEX_RULE_INDEP — R211–R214 HB-TASK-RULES (DARK, 2026-10-02). TB4.0
 # requirement-miss pass over the 16 never-passed tasks that fail ONE identical hidden test every run: an input field ignored or a given
@@ -1240,6 +1250,8 @@ CORTEX_ENDTURN_GATE=true
 # [LEDGER 2026-10-02 @4.124.43] DARK — BUILT + unit-tested (taskRules.test.ts: off = byte-identical planner + judge personas). NO bench arm yet. evidence: tb4-never-passed-2026-10-02 (cargo-flight-dispatch, data-anonymization, live-database-cutover).
 # [LEDGER 2026-10-04 @4.124.55 cell] FIRST REAL ARM (R211–R214 with R208–R210, cwbr vs cwb on MiMo-40, bundled base): +3 (20 vs 17),
 #   undetermined; engaged (task_rule_plan 40 / task_rule_judge 74); protected rejects 2 vs 3. TB4 validation run A is still the real test.
+# [LEDGER 2026-10-04 TB4 arm] R211–R214 with R208–R210 on TB4 (t4r): 3/11 vs c23 6/11 — no lift, negative-leaning (uefi-bootkit, risk-scorer-replay
+#   lost). Stays dark. Ledger r-tb4-valA-2026-10-04.md.
 #CORTEX_RULE_INPUTS=
 # R212 INTERP = L2 interpretation enumeration: named alternatives each computed; keep the one consistent with every explicit hint and all data.
 # [LEDGER 2026-10-02 @4.124.43] DARK — BUILT + unit-tested (taskRules.test.ts). NO bench arm yet. evidence: tb4-never-passed-2026-10-02 (foodstuff-beta-activity, glycan-ms2-elucidation, hof-topology-interpenetration, ontology-kg-querying).
@@ -1917,6 +1929,8 @@ MENTORSHIP_ACTIVE_DISCOVERY=false
 #   g6pro2 (Pro judge) 24 — best Luna cell, judge false-accept 38% vs 44%, 15 vetoes→PASS; fsppfj2 (deepseek-flash action) 23/40 vs
 #   rsysoff (Pro judge) 23 — WASH, false-accept 36% vs 42%. Both undetermined (|Δ|<4); replicate g6ppfj K=2 before claiming. Stays DARK.
 #   Ledger .cortex/bench/r-mimo64-ppfj-2026-10-03.md. (fsppfj VOID: harvest adjudicated the live run — fixed, 3 guards.)
+# [LEDGER 2026-10-04 TB4 arm] Flash judge on TB4 (t4j: c23 w stack + CORTEX_ENDTURN_RESOLVER_MODEL=deepseek-v4-flash, 14 c23 tasks): 6/14 vs
+#   c23 8/14 (−2, undetermined); 0 second attempts opened (its finishes are never soft accepts); $5.21. Stays dark. r-tb4-valA-2026-10-04.md.
 #CORTEX_LIFT_PLAN_MODEL=
 #CORTEX_ENDTURN_RESOLVER_MODEL=
 #CORTEX_DEADLINE_EXIT_MENTOR_MODEL=
@@ -2389,6 +2403,8 @@ CORTEX_ORIENT_BOOTSTRAP=1
 # SCOPE=true ("smallest set"), agent stashed 7 detector files, 1-file patch failed 10/10 hidden tests (only sglang run ever under SCOPE=true;
 # prior 12/20). DARK. Validate: MiMo regression (protected-reject rate must stay ~4%, not back to ~10%) + TB4 multi-file guards; then default.
 #CORTEX_ORIENT_BASELINE=
+# [LEDGER 2026-10-04 cells] =task first arms: MiMo cwbs 16 vs cwb 17 (−1, protected rejects 5 vs 3); TB4 t4s 2/11 vs c23 6/11 (halted at 11/14 by
+#   the key-wide budget cap) — no lift, negative-leaning. Stays dark. Ledgers r-mimo64-bundle-2026-10-04.md, r-tb4-valA-2026-10-04.md.
 #CORTEX_LIFT_PLAN_SCOPE=
 # [LEDGER 2026-09-02 @4.89.0] code default 30 (4.88.0) — 20 made the break rung (2N=24) unreachable; preserves 40%/80% density semantics.
 #NEARDUP_WINDOW=30

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nexus-cortex/cli",
-  "version": "4.124.56",
+  "version": "4.124.57",
   "description": "Nexus Cortex CLI - Terminal interface for multi-provider LLM orchestration",
   "main": "dist/index.js",
   "type": "module",
@@ -19,7 +19,7 @@
     "prepack": "node ../../scripts/copy-pkg-cortex-scaffold.mjs"
   },
   "dependencies": {
-    "@nexus-cortex/core": "4.124.56",
+    "@nexus-cortex/core": "4.124.57",
     "chalk": "^4.1.2",
     "cli-spinners": "^2.9.0",
     "commander": "^11.0.0",