@nexus-cortex/cli 4.124.16 → 4.124.17
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 +11 -0
- package/package.json +2 -2
package/.env.defaults
CHANGED
|
@@ -1172,6 +1172,17 @@ CORTEX_ENDTURN_GATE=true
|
|
|
1172
1172
|
# (no lever) R187c (4.124.15): the requirement-ledger extractor gets the planner's timeout (180 s, CORTEX_JUDGE_REQ_LEDGER_TIMEOUT_MS
|
|
1173
1173
|
# override) instead of the judge's 90 s (a 98-s pro call lost the whole ledger in the 9:04 AM L7 smoke), and an empty lift-time result is
|
|
1174
1174
|
# re-authored ONCE at the first finish.
|
|
1175
|
+
# (no lever) R191 HB-LEDGER-EXTRACTOR-HYGIENE (4.124.17, 2026-09-23; L7 interim read, 5 sessions): (a) the sglang ledger carried "The DOM
|
|
1176
|
+
# must not change" — copied from the persona's own example list — and the writer called it bizarre; (b) 4 of 5 ledgers spent 1–2 of their 8
|
|
1177
|
+
# lines on conduct text ("complete within 28800 seconds", "do not cheat / no online hints"); (c) photonic's extractor guessed a checker's CLI
|
|
1178
|
+
# (`check_routing.py <spec> <result>` vs the documented `--layout 1`), two lines FAILED at every finish on a PASSING solution, and the writer
|
|
1179
|
+
# spent turns on the "harness quirk". Fix: persona lists kinds abstractly, marks the example as from an unrelated task, skips conduct lines,
|
|
1180
|
+
# and invokes a named checker exactly as documented; the parser drops conduct lines (isBoilerplateRequirement); a check run whose failure is
|
|
1181
|
+
# the CHECK's own defect (usage/argparse error, missing script or tool, SyntaxError/NameError) is BROKEN → the line becomes UNVERIFIABLE with
|
|
1182
|
+
# its check dropped (never re-run, never a failure, Jev-decidable); the endturn_resolver event now banks reqBroken + reqLinesDetail (per-line
|
|
1183
|
+
# status/runs/last result) so the false-failure rate is readable from the decisions file. Interim L7 (5/21 rows): ledger authored + delivered
|
|
1184
|
+
# 5/5, 8 lines each, 1–6 with checks; reqOpen was 0 at every finish (the harness exercises every check itself, so the OPEN hold never fires;
|
|
1185
|
+
# the ledger's live effects are FAILED-as-evidence and Jev closes); every passing session had ≥1 FAILED line at its accepted finish.
|
|
1175
1186
|
# CORTEX_JUDGE_INDEPENDENT_DERIVATION / _TOL — R176 HB-INDEPENDENT-DERIVATION (4.122.0, 2026-09-21; design docs/R176_INDEPENDENT_DERIVATION_DESIGN.md).
|
|
1176
1187
|
# The largest never-passed TB4.0 class (~14 tasks: foodstuff, roy-polymorph, glycan, gsea, hof, vigenere, sound-change, cargo, ctr,
|
|
1177
1188
|
# shadow-relay, satb, takens…) is a computed value / exact artifact graded against a golden answer with NO ground truth in the container:
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@nexus-cortex/cli",
|
|
3
|
-
"version": "4.124.
|
|
3
|
+
"version": "4.124.17",
|
|
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": "4.124.
|
|
22
|
+
"@nexus-cortex/core": "4.124.17",
|
|
23
23
|
"chalk": "^4.1.2",
|
|
24
24
|
"cli-spinners": "^2.9.0",
|
|
25
25
|
"commander": "^11.0.0",
|