What's new in v9.0.10
- 1531c2a chore(release): bump version to 9.0.10
- 39ebc98 docs(readme): premium table Start here — 0→100 redesign of 5-door nav
- a8330e2 docs(readme): card-grid Start here + stacked install cards, no tables
- 0883412 docs(readme): premium redesign — card-grid Start here, polished sections
- a731a22 docs(readme): remove Core capabilities matrix from landing page
- 9b93dc1 docs(repo): move historical root reports into forensic archive
- 65ce483 refactor(live-engine): polish formatting and lint compliance for extracted seams
- 6778565 docs(repo): record branch and worktree retirement audit
- d9fec3e docs(provenance): add Agent/ routing guide and complete forensic taxonomy
- 760e9df docs(repo): consolidate forensic and agent provenance
- 42696a0 refactor(live-engine): extract WorkerSupervisor (Cluster 2)
- 8957cec refactor(live-engine): extract TemporalContract (Cluster 4)
- 17dd8c5 refactor(live-engine): extract LiveFreshnessService (Cluster 3)
- d995bd6 fix(ml): live engine online trainer + CLI train-model default off champion path (P0 fix follow-up)
- 66e9592 Nexus-Main: ruff lint+format heal for P0 producer wave (CI red at 811800a)
- 811800a docs(ml): PILOT_VALIDATION_RESULT + continuation report update — pilot passed, 34x10 justified
- fc36b2f test(ml): TEST-CURRENT-70D-16 updated for isolated candidate default (P0 follow-up)
- 3bd692c chore(lint): ruff fixes for P0 producer tests + pilot producer
- bc15c01 test(security): hot-swap governance — path allow-list / hash binding / eligibility gates (P0 step 8)
- 087bbc5 feat(ml): time-bounded pilot producer + full pilot-subset provenance (P0 step 7)
- d19195e fix(security): hot-swap governance + safe weights_only model loading (P0 step 6)
- 367d7f4 test(ml): P0 regression suite — exact producer failure modes pinned (step 5)
- 393f8ae fix(ml): atomic bundle publication in the canonical trainer (P0 step 4)
- 902ec0d fix(ml): typed dataset provenance binding for the trainer (P0 step 3)
- cf6a557 fix(ml): bind dataset provenance into trainer metadata + opt-in flag (P0 step 2)
- 4c4b007 fix(ml): producer champion-guard + isolated candidate output + hard emission gate (P0 step 1)
- 796e0bf docs(forensics): postmortem PARTIALLY_TRAINED + continuation + final model conclusion — 34x10 worker 25032 still hot
- f1ef7cb docs(forensics): final disposition — drop 6 audited stashes, Pages state, cleanup conclusion
- 345932e docs(forensics): recovery-kit 979624f — baseline + stash matrix + artifact/contract/readiness + security + main integration
- 29a8ebb Nexus-Main: ruff format heal on forensic_recovery_20260904/agent3-stash2-report.md (md code-fence line >88, mechanical)
- 979624f docs(forensics): 6 independent stash audits (one agent per stash)
- b015db9 ui(NSE): make Admin Dashboard responsive on phones/tablets/monitors/ultrawide (CSS only)
- a5e2ccc Nexus-Main: ruff format heal on scripts/docs/build_site.py (1-file drift after P3 cleanup)
- 694ee2b docs(forensics): reflect P3 completions (pushed 4261c3d)
- 4261c3d docs(site): regenerate site/_site (v9.0.9 drift heal + docs-enhance wiring)
- 54227f5 forensic(P3): wire docs-enhance assets in build_site.py shell + copy
- ebee9b8 forensic(P3): remove 600-line FLAG_BUILD_INDEX dead tail from build_site.py
- d3c59d4 docs(forensics): forensic recovery deliverables — stash triage + integration matrix
- 6bb7649 forensic: use compute_70d_frame_fast for 70d variants in three_model (BUG-106 extension)
- 706269f Nexus-Main: fix release.yml CRLF corruption — backslash+b (0x5c 0x62) mangled to backspace (0x08) in BUG-239 stage block (3 lines: Test-Path + 2x Copy-Item); heals 'Workflow static integrity scan' YAML parse failure at 62cfc51
Architecture support
- Windows x64: SUPPORTED (PyTorch/Polars/MetaTrader5 wheels exist)
- Windows ARM64: UNSUPPORTED — no wheels for the dependency stack
Tests ran in this workflow (gates job). No claims beyond that.