gstack/scripts/resolvers
Garry Tan 4328748136
fix(security): delete the dead ML layers — transcript classifier and DeBERTa ensemble
The L4b Haiku transcript classifier and the opt-in DeBERTa ensemble
(GSTACK_SECURITY_ENSEMBLE=deberta, a documented 721MB download) had ZERO
production callers since the chat-path agent that invoked them was ripped.
The only live ML path is scanPageContent (testsavant) inside the security
sidecar subprocess. Deleted by import graph:

- security-classifier.ts 614 -> 265 lines: HAIKU_MODEL, checkTranscript,
  shouldRunTranscriptCheck, loadDeberta, scanPageContentDeberta, ToolCallInput,
  all DEBERTA_* consts + load state. Header now states the live truth
  (imported only by security-sidecar-entry.ts). downloadFile kept, name
  intact — it is an enumerated egress sink (HF model download).
- security-bunnative.ts + test: a research skeleton self-described as 'NOT a
  production replacement', shipped into src/ with zero importers.
- security-bench-ensemble{,-live}.test.ts + the Haiku response fixture: a
  paid live-model benchmark for a layer that could not fire. The
  security-classifier-tdz test's only case exercised checkTranscript — gone.
- security.ts: layer-model header rewritten to the live architecture;
  StatusDetail.layers -> {testsavant, canary}; getStatus() no longer requires
  the impossible transcript==='ok' for 'protected' (old on-disk session state
  with a transcript key is tolerated on read, never re-emitted).
- security-sidecar-entry.ts needed zero changes: it serializes
  getClassifierStatus() verbatim and no consumer read .transcript (verified
  in sidecar-client + server.ts).
- BROWSER.md security section matches reality (ensemble knob gone, 112MB not
  22MB, sidecar hosting documented). combineVerdict/THRESHOLDS retained as
  the pure, tested combiner of record — comments now flag transcript/deberta
  votes as producer-less.

Net: 26 pass in security.test.ts incl. a NEW regression test for stale-
transcript disk tolerance; egress-receipt tripwire green.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-08-14 21:00:17 -07:00
..
preamble v1.62.0.0 feat: plan-mode auto-select at the review scope gate (#2533) 2026-08-12 11:12:28 -07:00
browse.ts fix: avoid tilde-in-assignment to silence Claude Code permission prompts (#993) 2026-04-16 14:49:56 -07:00
composition.ts v1.57.4.0 refactor(ethos): rename Boil the Lake principle to Boil the Ocean (#1912) 2026-06-08 05:41:07 -07:00
confidence.ts v1.43.2.0 fix wave: post-Daegu paper-cut — 18 fixes, 28 bisect commits (#1642) 2026-05-21 21:21:07 -07:00
constants.ts v1.57.10.0 feat: Codex review default-on across review/ship/plan/docs (#1966) 2026-06-10 21:14:58 -07:00
design.ts v1.45.0.0 feat(design): persistent board daemon — 24h boards, one tab, board history (#1710) 2026-05-25 20:45:12 -07:00
dx.ts feat: /plan-devex-review + /devex-review — DX review skills (v0.15.3.0) (#784) 2026-04-03 16:22:57 -07:00
gbrain.ts v1.52.1.0 feat: brain-aware planning — 5 skills read structured gbrain context before asking (#1742) 2026-05-29 08:35:00 -07:00
index.ts v1.57.10.0 feat: Codex review default-on across review/ship/plan/docs (#1966) 2026-06-10 21:14:58 -07:00
learnings.ts v1.33.1.0 fix(learnings): token-OR query + task-shaped retrieval in 3 long skills (#1442) 2026-05-11 19:34:33 -07:00
make-pdf.ts feat(v1.4.0.0): /make-pdf — markdown to publication-quality PDFs (#1086) 2026-04-20 13:20:30 +08:00
model-overlay.ts feat(v1.10.1.0): overlay efficacy harness + Opus 4.7 fanout nudge removal (#1166) 2026-04-23 18:42:58 -07:00
preamble.ts v1.58.5.0 feat: first-run activation scaffold + gstack router front door (#2078) 2026-06-25 09:42:45 -07:00
question-tuning.ts v1.61.0.0 fix wave: guards failing open / silent failures (9 fixes, 4 community PRs absorbed) (#2472) 2026-08-08 09:28:45 -07:00
redact-doc.ts v1.53.0.0 feat: smarter redaction — PII/secrets/legal guard across /spec, /ship, /cso, /document-* (#1797) 2026-05-30 08:54:46 -07:00
review-army.ts v1.42.0.0 Daegu wave: 23 community-filed bugs + PTY classifier enforcement (24 bisect commits) (#1594) 2026-05-20 07:35:01 -07:00
review.ts v1.57.10.0 feat: Codex review default-on across review/ship/plan/docs (#1966) 2026-06-10 21:14:58 -07:00
sections.ts v1.54.0.0 feat: carve /ship into skeleton + on-demand sections (-59% always-loaded) (#1806) 2026-05-30 12:09:10 -07:00
tasks-section.ts v1.38.1.0 fix wave: surrogate-safe page captures (#1440), Implementation Tasks across review skills (#1454), root-level artifact patterns (#1452) (#1504) 2026-05-14 21:46:50 -07:00
testing.ts feat(v1.3.0.0): open agents learnings + cross-model benchmark skill (#1040) 2026-04-19 17:50:31 +08:00
types.ts v1.46.0.0 feat: gstack v2 foundation — catalog tokens drop 56%, eval-first floor covers all 51 skills (#1712) 2026-05-26 16:50:03 -07:00
utility.ts feat(v1.5.2.0): Opus 4.7 migration — model overlay, voice, routing (#1117) 2026-04-22 01:06:22 -07:00