@nexus-cortex/cli 4.90.0 → 4.90.2

package/.env.example

@@ -4,7 +4,11 @@
 # ║ stands. Each lever carries a "# [LEDGER <date> @<version>] STATUS — evidence"      ║
 # ║ line (ON | DARK/RETIRED | DARK/PENDING | DARK by design). Update the line when a   ║
 # ║ cell moves the verdict and when each iteration ships. Cell ledgers live in        ║
-# ║ .cortex/bench/r-*.md; the k=5 candidate = ON set + the DARK/PENDING trio.          ║
+# ║ .cortex/bench/r-*.md. K=5 CONFIG (settled 2026-09-03 @4.90.1, r-v4-validation):    ║
+# ║ ON set + CORTEX_ENDTURN_GATE=true + CORTEX_ENDTURN_REQUIREMENTS=strict +           ║
+# ║ ENABLE_WEBTOOLS=false (bench pin) + VISION_HELPER_MODEL on + TOOL_TIMEOUT_MODE=auto; ║
+# ║ MENTOR_AUTO / EFFORT_TAIL / EFFORT_PULSE dark. Pin nexus-cortex@4.90.1 (4.90.0's    ║
+# ║ timeout-promote was inert). Both models.                                            ║
 # ╚══════════════════════════════════════════════════════════════════════════════════╝
 # Copy to .env and fill in the API keys for the providers you use. Every other value
 # has a proven-optimal default.
@@ -152,7 +156,7 @@ WEB_TOOLS_MODEL=gemini-2.5-flash
 #           {{toolNames}} prompt vars) AND refused at dispatch.
 # 🔴 Bench arms must PIN true/false explicitly: keyed bench containers carry job tokens for every
 #   provider, so auto resolves ON there. Hot-toggleable.
-# [LEDGER 2026-09-02 @4.89.0] ON=auto (NEW 4.89.0) — keyless A/B 08-31: wall pages + hallucination pressure; benches paid 3 taxes (pulse trigger fed by failing browse/WebFetch, Stage-5 unattested-web rejections, wasted turns). BENCH ARMS PIN false (keyed containers carry every provider token → auto=ON); c3 pinned false: 0 refusals, clean. Live: keyless 49 tools/WebFetch only; keyed 51/all; false 48/none.
+# [LEDGER 2026-09-03 @4.90.1] ON=auto (since 4.89.0) — v4 (bench pin false): 0 web refusals in either arm = the deny-list is clean and silent. — keyless A/B 08-31: wall pages + hallucination pressure; benches paid 3 taxes (pulse trigger fed by failing browse/WebFetch, Stage-5 unattested-web rejections, wasted turns). BENCH ARMS PIN false (keyed containers carry every provider token → auto=ON); c3 pinned false: 0 refusals, clean. Live: keyless 49 tools/WebFetch only; keyed 51/all; false 48/none.
 ENABLE_WEBTOOLS=auto
 
 # ============================================
@@ -256,8 +260,16 @@ MENTORSHIP_HELPER_MODEL=deepseek-v4-pro
 # and returns TEXT (the primary never receives image bytes — item 7's safety intent holds). Evidence: in EVERY
 # image task of the c3/p3 cells (chess-best-move ×2, gcode-to-text, code-from-image ×2) the text model SEARCHED
 # for an image tool by name, found none, and hand-built tesseract/PIL pipelines that ate the budget. Same DeepSeek
-# key. Set false to restrict ReadImage to vision-capable primaries. LIVE 09-02: flash transcribed a test PNG
-# exactly (3 lines, 1 iteration) through the helper; bench cell on chess/gcode/code-from-image pending.
+# key. Set false to restrict ReadImage to vision-capable primaries.
+# [LEDGER 2026-09-03 @4.90.1] ON — VALIDATED on the real bench (r-v4-validation-2026-09-03, n=1): fired HEAVILY
+# (flash 25 / pro 45 vision_handoff events) and drove flips — flash chess-best-move PASS (ReadImage×6, first flash-family
+# chess pass ever), pro dna-insert PASS (vis×6, p3-without-vision failed it), code-from-image PASS both models. Headline
+# k=5 lever. POST-RUN FINDINGS (audit of the 9 fails): (a) NO per-turn hand-off budget — gcode-to-text burned its whole
+# task budget on self-rendered slice reads (flash 14 / pro 34 ReadImage calls, never converged) → planned lever
+# VISION_HANDOFF_MAX (~8/turn) + a "consolidate into ONE labeled montage" reminder (def-fcb719427f); (b) structured
+# images misread — pro chess got a wrong whole-board transcription → bad FEN → nonsense move, while flash passed via a
+# montage of individually LABELED squares (def-a3e7920cf9): grid/crop-aware prompts beat whole-image reads; a stronger
+# vision card is the next lever once available.
 VISION_HELPER_MODEL=deepseek-v4-flash-vision-exp
 # Periodic mentorship review every N turns
 MENTORSHIP_TURN_BASED_ENABLED=false
@@ -330,9 +342,9 @@ TOOL_BUDGET_SOFT=400
 # Per-tool execution timeout in milliseconds (2 minutes). With TOOL_TIMEOUT_MODE=auto this is the
 # PROMOTE threshold in headless sessions (the command keeps running in the background), the KILL
 # deadline in interactive ones.
-# [LEDGER 2026-09-02 @4.90.0] ON=120000 — p3 timeout audit (r-p3): the 120s kill fired 42× across 14 task-runs and RESCUED NONE (train-fasttext busy-wait driver, long builds); every p3 timeout was Harbor's budget, not ours. Kept at 120s because TOOL_TIMEOUT_MODE=auto now promotes instead of kills in headless (item 8); v4 cell measures tool_promoted events.
+# [LEDGER 2026-09-02 @4.90.0] ON=120000 — p3 timeout audit (r-p3): the 120s kill fired 42× across 14 task-runs and RESCUED NONE (train-fasttext busy-wait driver, long builds); every p3 timeout was Harbor's budget, not ours. Kept at 120s because TOOL_TIMEOUT_MODE=auto promotes instead of kills in headless (item 8). v4 audit: pro/train-fasttext = the busy-wait driver again (249 bash, 22 cancels) and 4.90.0's promote was INERT (outer-abort race, fixed 4.90.1) → train-fasttext is the 4.90.1 re-test task.
 TOOL_TIMEOUT_MS=120000
-# [LEDGER 2026-09-02 @4.90.0] ON=auto (NEW) — PROMOTE-AT-DEADLINE: when a Bash call exceeds TOOL_TIMEOUT_MS in a
+# [LEDGER 2026-09-03 @4.90.1] ON=auto — 🔴 4.90.0 was INERT: the orchestrator's outer abort raced ShellTool's deadline and always won (v4 pro/train-fasttext: 22 cancelled, 0 promoted). 4.90.1: outer cap = TOOL_TIMEOUT_MS+30s, ShellTool default follows TOOL_TIMEOUT_MS; LIVE-VERIFIED (5s deadline, 9s command → tool_promoted event, model read the background output). Re-test task: train-fasttext. PROMOTE-AT-DEADLINE: when a Bash call exceeds TOOL_TIMEOUT_MS in a
 # headless/auto-approve session, the still-running command is moved to a background shell (bash_id) and the tool
 # returns the output so far instead of killing it; interactive sessions keep the kill. Evidence (c3/p3 trajectories):
 # the 120s kill fired 42x across 14 task-runs (19 legitimate scripts, 12 wait loops, 5 builds/trainings) and rescued
@@ -435,7 +447,7 @@ ROUTER_MIN_SAMPLES=3
 
 # Mandatory EndTurn pre-delivery self-audit + Stage 2/3 verifiers. Default OFF; ON =
 # graded cortex-channel training records (the tool is hidden when off).
-# [LEDGER 2026-09-02 @4.89.0] DARK/PENDING → **RECOMMENDED ON for k=5** — 3 cells: 4.71 fp 11→1 · fd +2/5 · c3 (n=2, r-c3-candidate-config) code-from-image FLIP 2/2 (first flash-family pass ever; Stage-4 requirement rejections, no mentor) + dna-insert 1/2; 0 clean regressions outside its LIMIT (accepts a re-attested wrong artifact: pytorch-model-cli r2, dna-insert fd). Cost: attestation turns. Operator decision pending. PRO (r-p3, n=1): controls 3/3 held with NO attestation tax (bn-fit 53→30 turns), 0 regressions, 0 paralysis; 0/13 flips on pro's hard core (budget timeouts while working). → k=5 config for BOTH models: guards + GATE+REQ + ENABLE_WEBTOOLS=false, mentor/tail off.
+# [LEDGER 2026-09-03 @4.90.2] DARK/PENDING → **RECOMMENDED ON for k=5 (with REQUIREMENTS=strict)** — COST (v4 pass mining + doctrine-mine): every pass paid 4-8 EndTurn round-trips, the biggest share = Stage-2 CITATION rejects (17/43 doctrine findings; the model never re-ran a command for a fresh verbatim line) → 4.90.2 nudge teaches per-turn grounding + re-run; expect controls' +25-35% turn tax to drop — measure at k=5. — v4 (n=1, @4.90.0): gate engaged heavily under strict (flash gate_req=22/cite=13, pro 17/4), controls bn-fit+sqlite held on BOTH models, pro flips vs p3: dna-insert, pytorch-model-cli, cancel-async (n=1 candidates). Prior 3 cells: 4.71 fp 11→1 · fd +2/5 · c3 (n=2, r-c3-candidate-config) code-from-image FLIP 2/2 (first flash-family pass ever; Stage-4 requirement rejections, no mentor) + dna-insert 1/2; 0 clean regressions outside its LIMIT (accepts a re-attested wrong artifact: pytorch-model-cli r2, dna-insert fd). Cost: attestation turns. Operator decision pending. PRO (r-p3, n=1): controls 3/3 held with NO attestation tax (bn-fit 53→30 turns), 0 regressions, 0 paralysis; 0/13 flips on pro's hard core (budget timeouts while working). → k=5 config for BOTH models: guards + GATE+REQ + ENABLE_WEBTOOLS=false, mentor/tail off.
 CORTEX_ENDTURN_GATE=false
 # Stage 4 (4.70.0, requires the gate ON): task-requirements attestation —
 # enumerate each stated requirement with what satisfies it + how it was
@@ -446,7 +458,7 @@ CORTEX_ENDTURN_GATE=false
 # (a claim with no matching tool output is rejected). Motivation: 4 pro fails
 # (cancel-async-tasks, dna-assembly, dna-insert, pytorch-model-cli) were 85-95% complete with
 # TRUE attestations verified against the model's own tests, not the task's exact constraints.
-# [LEDGER 2026-09-02 @4.90.0] DARK/PENDING → `true` recommended ON with the gate (c3: code-from-image flip ×2; p3: safe on pro, 0 tax); `strict` = v4 validation cell (fresh, unmeasured). Needs GATE.
+# [LEDGER 2026-09-03 @4.90.1] DARK/PENDING → `strict` RECOMMENDED for k=5 with the gate. v4 cell (n=1): strict rejects fired heavily (flash gate_req=22, pro=17), 0 control regressions. 🔴 4.90.0 strict CRASHED on EndTurn({}) (null.filter ×4, flash/dna-insert) and rejected 11 honest condensations as paraphrases (protein-assembly) → 4.90.1: null-guarded; verbatim = 24-char window OR ≥70% content tokens; nudge shows a pasted-output example. Needs GATE.
 #CORTEX_ENDTURN_REQUIREMENTS=false
 
 # ============================================

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nexus-cortex/cli",
-  "version": "4.90.0",
+  "version": "4.90.2",
   "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": "4.90.0",
+    "@nexus-cortex/core": "4.90.2",
     "chalk": "^4.1.2",
     "cli-spinners": "^2.9.0",
     "commander": "^11.0.0",