@nexus-cortex/cli 4.125.3 → 4.125.4
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.env.defaults +7 -0
- package/package.json +2 -2
package/.env.defaults
CHANGED
|
@@ -296,6 +296,13 @@ CORTEX_EFFORT_TAIL=false
|
|
|
296
296
|
# [LEDGER 2026-09-11 @4.107.0] cell-n-r4 (hard core 13×K=2, flash action model, FINAL 47/52 rows): high = NO LIFT — ah-none 6/22 (0.27), ah-low 7/25 (0.28) vs control 11/26 (0.42); +3.5–4.5 min/task latency, turns unchanged (narrow door intact). Keep unset (card default). DARK. Ledger r-cell-n-r34.
|
|
297
297
|
# [LEDGER 2026-09-12 @4.107.2] PRO @high on the k=5 series COMMON (tb21-k5-ph1, full 89, K=1): 72/89 = 80.9% (Wilson 0.72-0.88), $12.77, med 737 s — +3-4 pts over the flash series (77.3 ± 2.7), CIs overlap; effort NOT separated from model (pro @default not yet run). Flash @high = no lift (cell-n-r4). Stays DARK. Ledger .cortex/bench/r-tb21-k5-ph1-2026-09-12.md
|
|
298
298
|
# [LEDGER 2026-09-18 @4.119.1] TB4.0 cell e6 (29 ever-passed tasks, 3-h budgets, flash, 1 repeat): LOW = REJECT 3/29 vs high 10/28 (+2/−9, p 0.07; tokens equal, wall +60%); MAX = WASH vs high 10/29 (+4/−4) at +30% tokens/+60% wall. Wire: reasoning/iteration FLAT across levels (4.6K/5.0K/4.8K chars), only the peak call scales (15.8K/22.7K/26.1K tok) — effort caps peak reasoning, not typical. Keep HIGH for TB4.0. Ledger r-tb4-e6-effort-2026-09-18.md.
|
|
299
|
+
# [LEDGER 2026-10-17 @4.125.4] `none` = the action model runs with THINKING OFF (the non-reasoning arm). Live probe on deepseek-flash (1017):
|
|
300
|
+
# {thinking:enabled + reasoning_effort:none} still reasons; {reasoning_effort:none} alone and {thinking:{type:disabled}} do not. Until 4.125.4 the
|
|
301
|
+
# harness OMITTED reasoning_effort for `none` and sent no thinking field → the provider default (thinking ON) — a `none` arm would have been a
|
|
302
|
+
# silent control. Now APIClient sends thinking.type=disabled on every DeepSeek chat request with effort none (same field the forced-tool_choice
|
|
303
|
+
# path uses). Wire proof per row: reasoningTokens == 0 / reasoning_share == 0. History: effort LOW rejected (e6 09-18: 3/29 vs 10/29); no
|
|
304
|
+
# thinking-OFF action arm has ever run — hypothesis: no reasoning walls / degenerate loops (a thinking-mode phenomenon, ~20% of failures) and
|
|
305
|
+
# forced tool calls everywhere, at the cost of raw capability on the deciding calls. First test = 14-task subset vs arm C of t4w1.
|
|
299
306
|
#CORTEX_ACTION_EFFORT=
|
|
300
307
|
# Trigger helper-model review on tool errors (only applies when mentorship is enabled)
|
|
301
308
|
MENTORSHIP_TRIGGER_ON_ERROR=false
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@nexus-cortex/cli",
|
|
3
|
-
"version": "4.125.
|
|
3
|
+
"version": "4.125.4",
|
|
4
4
|
"description": "Nexus Cortex CLI - Terminal interface for multi-provider LLM orchestration",
|
|
5
5
|
"main": "dist/index.js",
|
|
6
6
|
"type": "module",
|
|
@@ -19,7 +19,7 @@
|
|
|
19
19
|
"prepack": "node ../../scripts/copy-pkg-cortex-scaffold.mjs"
|
|
20
20
|
},
|
|
21
21
|
"dependencies": {
|
|
22
|
-
"@nexus-cortex/core": "4.125.
|
|
22
|
+
"@nexus-cortex/core": "4.125.4",
|
|
23
23
|
"chalk": "^4.1.2",
|
|
24
24
|
"cli-spinners": "^2.9.0",
|
|
25
25
|
"commander": "^11.0.0",
|