v9.0.11

What's new in v9.0.11

• 5f80448 test(disk): commit the three manifest-referenced batteries missed by ff7043b
• 63e0a2a chore(release): regen dependency lock for 9.0.11 (ast-serialize 0.11.1 drift)
• ff7043b fix(disk): never leak disk from the repo core — core-dump guard + sqlite URI contract
• bb249a5 chore(release): bump version to 9.0.11
• 9e55081 test(flush): drop racy still-queued assertion — worker may drain before the pre-flush read (CI #1014)
• f15b53b fix(gate): provisioned bundle ships scaler+meta — L4 artifact width split checkpoint=70 scaler=None (CI #1010)
• 565614d fix(ci): dependency drift resolver pins python-version 3.11 + regen lock for ruff 0.16.6
• d77ec79 NSE-Swarm QA/Tests: SMOKE-COUNT-CI closed via PR #114 (ci-e2e-fix branch + merge de9ce8c); deplock drift was runner-side (uv input misnamed + no 3.11); critical_coverage flagged as next P1
• 24ad92a fix(security): close OSV false positives #1075-#1080 — commit uv.lock + --no-resolve in osv-scanner
• b1ab401 fix(ci): setup-uv@v7 input is uv-version (renamed by PR #109) — old version key is silently ignored again
• 285769d fix(deps): regenerate requirements.lock against pyproject resolution (resolves lock staleness flagged by dependency-lock gate)
• 7d3e750 fix(ci): drift resolver must see python 3.11 — ubuntu-24.04 runner has none, uv pruned <3.12 marker arms (numpy 2.4.6/tomli) and drifted the committed universal lock
• de9ce8c fix(ci): nightly client E2E - playwright extra, E2E_BASE env, compose auth token (#114)
• c6cd03a chore(deps): bump ruff from 0.16.5 to 0.16.6 (#111)
• 8811c66 chore(deps): bump softprops/action-gh-release from 2.6.2 to 3.0.3 (#113)
• 91f32e6 chore(deps): bump actions/cache from 4 to 6 (#112)
• cd0d609 chore(deps): bump actions/upload-pages-artifact from 3 to 5 (#110)
• c886af2 chore(deps): bump astral-sh/setup-uv from 5 to 7 (#109)
• 28be277 chore(deps): bump actions/upload-artifact from 4 to 7 (#108)
• c7f9f46 ci(debug): temporary drift-debug workflow — dump CI fresh-lock numpy/polars arms (root-cause in-flight)
• 9fdacfb fix(gateway): stop exposing exception/stack-trace text to HTTP clients (CodeQL py/stack-trace-exposure)
• deb002a fix(ci): setup-uv@v5 input is version, not uv-version — pin was ignored, CI ran uv 0.12.11 and drifted the lock gate
• cbf663d fix(ci): nightly client E2E - generate throwaway NSE_WEB_AUTH_TOKEN for compose core
• c462e61 fix(ci): nightly client E2E — playwright extra + E2E_BASE env honored in journeys
• b461768 NSE-Swarm QA/Tests: smoke-count CI deeper root cause — scaler-less provisioning kills gate L4; full-bundle provisioning verified (runtime_gate CERTIFIED, smoke_self 13 passed)
• 4a67e09 fix(config): exit-policy keys ride the runtime snapshot — hot-reload/restart no longer reverts operator overrides
• f788b97 coordinator: wave-3 CI-recovery rows — CI-RECOVERY-987 / EXIT-SEPARATION-DELIVERY / LEDGER-EXPORT status (append-only)
• e5c7162 NSE-Swarm QA/Tests: swarm log entry — smoke-count CI root-cause probes + deplock diagnosis (diagnosis-only run)
• cea35e0 test(audit): export validator for TASK-EXIT-LEDGER-EXPORT (read-only, synthetic fixtures test the validator only)
• 5ba5675 fix(smoke): provisioned artifact now ships matching manifest.json digest
• 81bb5b2 fix(tests): smoke chain dispatch deterministic vs maintenance-window wall clock (CI #987)
• ba910f7 feat(exit): TASK-EXIT-SEPARATION — configurable exit policy, AI-flip suspended by default
• e2ec524 feat(policy): XAUUSD-only symbol whitelist gate — wave-3 operator ruling WIP landing
• abeb80b fix(ci): mkdir ci-results/coverage before critical-gate redirect (run #987 rc=1 No such file) + pin uv 0.12.10 for byte-reproducible lock
• 1f7c982 🔥 remove probe
• 9dfe1eb probe: workflows write permission test
• b104367 style: ruff format tick_pipeline.py — BUG-253 hunk reflow (CI #987 ruff_format direct fix)
• 47ad59b style: ruff format sweep — 10 committed files unformatted (CI #987 ruff_format class; format-only, no logic change)
• 0630971 deps: regenerate requirements.lock (ast-serialize 0.10.0->0.11.0) to clear dependency-drift gate
• e5fa529 coordinator: P0-1..P0-6 dispatch handoffs + taskboard row (deep audit Section K)
• 76a3da0 Nexus-QA-REPLAY: counterfactual exit-policy replay harness (wave 3, honest bounds model)
• 305adf3 Nexus-Coder-A4: A4 dead-letter store split — AuditRepository facade delegation, zero schema change (audit rev2)
• bd95636 coordinator: C4 row closed (8c2360b)
• d584871 coordinator: wave-3 taskboard claims — EXITREPLAY-HARNESS / EXIT-SEPARATION / LEDGER-EXPORT (user exit-behavior brief)
• 8c2360b Nexus-Coder-C4: C4 policy symbol/magic hardcode removed — multi-symbol unlock (audit rev2)
• 19c4aa4 coordinator: HOUSEKEEPING row closed (a77f831 + b6f0624)
• b6f0624 coordinator: fix docker_08 env-secret test over-broad pattern — commented placeholder falsely flagged (audit wave-2)
• a77f831 Nexus-DevOps-HK: I6 scratch/ untracked + B4 SHA256/SmartScreen docs + F9 halt-semantics invariant doc (audit rev2)
• cd2d298 coordinator: H5 taskboard row closed (e637b9f)
• e637b9f coordinator: H5 candle-intelligence honors AppConfig — hardcoded enabled=True defeated the disable ruling (audit rev2)
• 2950026 coordinator: wave-2 taskboard claims — C4/A4/H5/HOUSEKEEPING
• a192762 coordinator: audit wave-1 taskboard ledger — G1/G3/I9/C3/E1 commits + statuses
• 80f17ef Nexus-Coder-E1: E1/E2 zero-friction guard + canonical cost defaults in all research engines (audit rev2)
• 59e6473 coordinator: fix BUG-229-suite fixture drift — zone-gate test reachable again (audit wave)
• 76eb23b Nexus-Coder-C3: C3 spread percent-of-TP + session-percentile gates (audit rev2)
• f462fe5 Nexus-QA-G3: I9 HIST-SIM v1 golden fixture test (audit rev2)
• 15d3955 Nexus-QA-G3: G3 hold-age wall-clock fallback removed — conservative 0.0 + loud WARNING (audit rev2)
• a178c7c Nexus-Coder-G1: G1 broker UTC-offset configurable + detection helper (audit rev2)
• 3d127dc fix(live): BUG-253 duplicate process_tick clobbered is_new_bar -> stale liquidity blocked 70D inference
• 71c557d fix(ci): run #985 — WAL read-back poll in flush race test
• 2888ab8 fix(ci): run #983 three failures — deterministic flush race, artifact provisioning, web-auth parity
• 0e82f19 fix(ci): run #982 follow-ups — pynacl dependency + web-auth test adaptation
• 5ee6a72 fix(types): mypy 96->0 across seam modules + real async bug found in bootstrap-train (CI run #981)
• b067535 chore(deps): regenerate requirements.lock — ast-serialize 0.9.0 -> 0.10.0 (drift gate, run #979)
• 53b5de3 fix(qa): mutation catalog anchor drift — MUT-SHADOW-BOUNDARY anchor no longer matches formatted source (CI run #978/#979)
• 4a3c487 style: ruff lint fixes + canonical format sweep — unblock CI quality gate (run #978)
• 9b57174 docs(agent_handoffs): audit findings rev 2 — re-verification after agent fix waves + security P0 closures
• 4ab19dc chore(compose): wire NSE_WEB_AUTH_TOKEN as required env for core (audit B1 close-out)
• fec8e63 fix(gateway): fail-closed credential resolution — no silent default secrets (audit B2)
• 166fac1 fix(compose): postgres no host bind + mandatory password (audit B3)
• 7c14451 feat(web): WEB-AUTH-P0 token auth on the control-plane server (audit B1)
• f2a5b8c docs(agent_handoffs): independent 10-wave audit findings + fix status matrix (A-K, ~60 findings, P0-P3 dispatch order)
• 302ffdc chore(scratch): sync tracked research artifacts (bench comment fix, LE head snapshot, pytest run logs)
• e1faf60 docs(agent_handoffs): record 2026-09-07 absorption disclosures and integration-recovery handoff
• 76713f1 fix(live): repair broken seam extractions crashing the production tick path
• 400249f fix(lifecycle): GATE4/GATE5 consume the honest NOT_AVAILABLE sentinel fail-closed (ECON closure P0)
• 12b7ae9 docs(taskboard): register TASK-RESEARCH-TRAINING-PARITY-20260907 — walk-forward/replay/determinism/economic-objective mission COMPLETE
• 5fe7955 style(release-test): import sort only — unblock BeforePush ruff gate (no behavior change)
• f1c2c14 feat(execution): enforce canonical maintenance-window guard on live entry dispatch (ECON v1 phase 6, P1)
• 49a551c test(learning): restart-recovery + LiveEngine handoff contract tests (learning-loop P1)
• d8581e1 test(training): calibration-degradation gate contract tests (P1 follow-through)
• 64d1840 feat(lifecycle): final-replica training — training variance measured, median artifact selected (P2)
• 49916d5 feat(live): wire LearningCycleOrchestrator into LiveEngine behind config (learning-loop P1)
• 6fd8024 docs(taskboard): register TASK-LOOP-VALIDATION-20260907 — loop runs, candidates honestly rejected, no edge promoted
• c76cc1f fix(factory): double-encoded config crashes build_memory on loop restart — evolution memory coerced safely
• bfb8c08 test(expanding): patch the orphaned top-level WalkForwardTrainer class too
• 8d09abc style(taskboard): whitespace-normalize taskboard row (LF, no trailing spaces)
• 264ff26 feat(marketplace): measurement backtest migrates off zero-friction legacy default (ECON v1)
• 49af8a8 docs(taskboard): register TASK-RUNTIME-SAFETY-P0 — persisted halt/kill, audit dead-letter, queue criticality, error circuit, account freshness: COMPLETE
• ac414e8 feat(lifecycle): REAL fold metrics in TrainingRun.metrics (learning-loop P1)
• 60d9a11 test(expanding): robust _fit_scaler spy across duplicate module instances + calibration-degradation gate wiring
• dc5cceb feat(web): runtime-safety observability surface — persisted state, error circuit, account freshness + data-integrity counters in debug_snapshot risk section
• ebcf6a1 docs(audit): final production-readiness audit — evidence matrix, 2 audit fixes, honest-status verification (TASK-FINAL-AUDIT)
• b667ce8 style(econ): ruff residual cleanup — unused dataclass import, blank lines
• 2f77153 test(lifecycle): champion bundle recovery contract + 70D serving-trust regression
• 56fce1e feat(observability): GET /api/operator/calibration — read-only calibration evidence monitor (collection mission)
• bb5305e style(cli): quote risk_commands handle annotation (F821 forward-ref) — ruff clean
• 35726a1 feat(cli): nexus risk status|release — smallest safe explicit release surface for persisted safety states (P0)
• 7ad8305 fix(factory): survive double-encoded generation config — completion no longer crashes before persisting COMPLETED
• ab9db74 fix(test): align launcher fixtures with artifact integrity contract
• 1d101bb feat(training): economic fold metric — net expectancy in R, honestly labeled
• 466e153 fix(live): complete L14 seam residue in live_engine — missing imports + staticmethod/classmethod decorators
• f8ef51a fix(live): re-apply hot-path consecutive-error circuit breaker on the L14 pipeline seam
• 825a261 docs(taskboard): ABSORPTION-DISCLOSURE-3F8F6206 — FACTORY kick carried in calendar commit, provenance preserved
• fd94050 feat(factory): prime the autonomous-loop pump on /loop/start + tick reentrancy guard (factory-loop audit fix)
• b4a79b2 style(live): ruff conform live_engine (F401 + format)
• 7828edd chore(economics): lint conformance pass — typing import, zipped strict, ruff fixes
• f729225 fix(live): IMT5Port annotation import in runtime_mode (TYPE_CHECKING) + format
• 3f8f620 fix(calendar): compose CalendarWorker into the live maintenance cycle (P0 Phase 4 — event gate becomes reachable)
• af51bf7 P1 seam L13: execution-mode transitions leave live_engine
• 549e980 fix(live): model_health Path/ScalpNet/ModelBundle name imports (residue completion)
• 8fcf6cb fix(live): import polars/numpy/torch in model_health (extraction residue F821)
• 6946f30 style(risk): ruff conform runtime-safety files — import sort + format only
• 3b6702c fix(live): champion_sync evaluate is a @staticmethod — N805 + LiveEngine TYPE_CHECKING import + format
• 4b30398 P1 seam L12: model-health pair leaves live_engine
• 4fffea5 feat(governance): wire dormant production-eligibility + collapse gates into controlled training (promotion audit follow-through)
• 27d0ac5 P1 seam L11: champion-registry sync leaves live_engine
• faacc29 style(execution): ruff conform hold_duration (RUF002 ambiguous minus + format)
• 72875d9 feat(risk): sizing policy confidence factor follows the calibrated-confidence contract (ECON v1 alignment)
• 00a821b feat(live): account snapshot freshness gate — stale/failed account refreshes block sizing decisions (P1)
• 45f4260 style(live): ruff conform live_engine (F401 + format) — L-wave continuing repair
• df1ffc2 feat(calibration): identity-bound calibration + evidence collector — artifact stays OFF until real OOS evidence exists (blocker resolution: SAFE-BLOCKER PATH)
• b6c6894 P1 seam L10: warmup cluster implementation module (completes live_engine delegates)
• 1b30d0c style(live): ruff conform tick_pipeline (F401 + format)
• bc49dd4 docs(governance): ABSORPTION-DISCLOSURE-792D1A38 — foreign runtime-safety WIP carried in bench-correction commit, provenance preserved
• 792d1a3 docs(taskboard): benchmark arm-semantics correction — 60D arm is a dead-input ablation, not canonical scalp_v2
• 3ca9bbe fix(live): model_bundle_store name repairs — TYPE_CHECKING bundle types, engine-arg fix, LiveEngine late import
• 7fca459 test(research): backtest <-> canonical economics parity harness (ECON v1 phase 9)
• 2422c7e style(live): ruff conform model_bundle_store (I001 + format)
• 68391f1 feat(execution): enrich canonical cost spec from the full cost-site forensic (PHASE 2B)
• 9ca824d feat(economics): server-offset correction + rollover/swap + maintenance policy tests (ECON v1 phase 6)
• d5a0142 style(adapters): ruff conform paper replay_source (F401 + format)
• 0752aaa style(adapters): ruff conform audit_repository (I001 import sort + format)
• 7a92a76 feat(installer): resumable 3-model matrix — train_all --resume skips variants with PASS evidence (first-run GAP 3)
• ac0f9b7 feat(risk): ConcurrencyPolicy — position-count vs total-exposure separation (ECON v1 phase 5)
• e7952f6 style(research): zip strict=False in block importance weight map (ruff B905)
• c0624f0 docs(taskboard): TASK-MARKET-CONTEXT P0 wave closure row — calendar/health/provenance/budget/drift/candle decision/block importance
• 7cecc86 feat(installer): nexus data-fetch — first CLI producer for data/raw historical bars (first-run GAP 2)
• 121e93d feat(observability): news-LLM budget + calendar/event-gate legs on /api/news/health (P0 Phase 10)
• acf294b docs(repo): record 2026-09-07 hygiene pass — stale TEMP worktrees + merged branches retired with WIP preservation
• 20dc231 docs(taskboard): register TASK-QA-HARDEN-20260907 - QA/CI hardening mission complete
• 3177f36 ci(local): pre-push gate gains critical-manifest drift stage (gate parity with CI)
• 16b74b1 feat(research): reproducible feature-block importance harness (P0 4A)
• b34f9de docs(taskboard): register TASK-PROMSTAT-P0P1P3 — statistical promotion gate + DST sessions + decision-ID integrity evidence
• 555e6df feat(installer): doctor DATA check — zero-data machines can no longer read READY (first-run honesty)
• 5a717cb feat(training): session-semantics provenance stamp on training metadata (P1 completion)
• 0fa6be8 feat(telegram): wire the command bus into the engine lifecycle — fail-closed construction, runtime-scoped start
• f482c9c ci(nightly): chaos job also runs the update-lifecycle scenario battery
• a373ee2 test(updater): lifecycle/chaos scenario battery - 11 scenarios over the real update engine
• ced18ba fix(candle): disable unused candle-intelligence runtime (P0 Phase 3 — Option B)
• c1f22d1 fix(model-generation): deterministic dataset rebuild is idempotent — intact same-content artifact is REUSED, not overwritten
• 0431c9e feat(training): thread REAL per-trade MAE/MFE from the ledger into gate evidence
• 664078a feat(news): dedicated scoped news-LLM budget (P0 Phase 5)
• 37c2589 P1 seam L9: model-bundle load/verify/persist cluster leaves live_engine
• fc552e3 docs(taskboard): ABSORPTION-DISCLOSURE-130AB14F — decision-ID files carried in P0 write-uncertainty commit, provenance preserved
• 9b02a8e fix(governance): register the retrain candidate before the lifecycle transition — ends the silent registry no-op
• 6bb0313 style(features): ruff conform session_time (all sort + formatter)
• 62be80e fix(application): persisted safety state restore-first at boot — HALTED/KILL_SWITCH survive restart and refuse trading until explicit release
• f99c213 feat(governance): economic-integrity promotion gate — fail-closed provenance keys (ECON v1)
• 9913f9f test(promotion): policy SSOT pins for the statistical gate + cfg-instance passthrough
• 26c5d9e ci(e2e): nightly client E2E - isolated compose core, readiness polling, failure artifacts
• 0590d6a fix(training): persist RUNNING training-run row before training begins — real restart-safety
• fe2e96e ci(nightly): scheduled deep-QA tier - slow suite, lifecycle chaos, full mutation campaign
• b281995 feat(news): deterministic high-impact LLM-skip in PRO cycle (P0 5A/5B)
• f2c8abd ci(qa): enforce critical-manifest drift gate + mutation-runner smoke + per-file coverage floors on push
• b57a5ca docs(taskboard): register TASK-REDIS-ISOLATION — redis profile-gated, default startup barrier removed (companion row to 92e0620)
• b9886de docs(taskboard): register TASK-ORDER-WRITE-UNCERTAINTY — typed write tri-state + intent store delivered (companion row to 130ab14)
• 1d2e98b docs(taskboard): register TASK-MKTCAL-ENTRYGATE — canonical MarketEntryGate delivered (companion row to a8aeb35)
• bf580a5 docs(taskboard): register TASK-MONEYPATH-CLOSURE-20260907 — evidence + honest gaps
• fae62aa docs(release): P3 artifact trust-chain contract — code-signing readiness gate + evidence-graded ONNX verdict
• 459ebbe test(live): convert SHADOW-boundary source coupling to ownership contract (L2 decision_executor)
• 1597184 fix(news): evidence-based source triage — dead sources handled, 2 live replacements (P0 1C)
• 88a8627 feat(installer): doctor EXECUTION_COSTS check — canonical cost artifact gates first-run green (PHASE 9)
• b1ece04 feat(lifecycle): P1 fine-tune atomic activation — integrity sidecars rebind to new weights before the serving swap
• 130ab14 feat(execution): order-write uncertainty semantics — typed WriteResult tri-state, UNKNOWN != FAILED, fingerprint reconciliation + restart-safe OrderIntentStore (P0)
• a8aeb35 feat(execution): canonical MarketEntryGate — broker-clock market state + weekly open-guard + Friday cutoff as a reusable entry decision (P1)
• 248507e style(governance): flatten nested min in feature_drift (ruff PLW3301)
• e1288be feat(research): BacktestEngine imports compute_sized_economic_pnl (ECON v1 wiring completion)
• 3c951d2 feat(risk): feature-drift circuit breaker state — canonical shadow70 drift monitor wired to the serving scaler (data-only)
• d9f6b57 feat(training): bounded deterministic anti-forgetting replay policy (PHASE 5)
• 30f4d8b test(qa): mutation-runner smoke tests - resolver contract, catalog integrity, arbitrary-cwd invocation
• b64aa3f test(market-context): 27 focused tests — calendar contracts, providers, gate, worker, drift (P0 12)
• 9c6e74a feat(features): DST-aware London/NY session semantics + artifact revalidation gate (P1)
• 2bf63d4 feat(shadow): statistically defensible promotion gate — bootstrap CI on the paired delta (P0)
• 23ce1aa feat(governance): paired moving-block bootstrap statistical promotion gate (PHASE 7B)
• 7e7dd1b test(mutation): hybrid mutation catalog - domain-contract mutations + EOL-tolerant matching
• d785cae feat(news): PSI drift monitor for the news 10D family (P0 2C)
• 7b9e132 feat(training): time-based purge/embargo option for walk-forward folds
• 8759811 feat(training): idempotent dataset-state retrain trigger + challenger-vs-champion comparison wired (learning-loop closure)
• a4388ae feat(execution): monotonic HoldDurationClock — DST/NTP-immune holding-duration semantics with sanity divergence forensics (P1)
• 37072a8 feat(release): P0 signed update trust root — Ed25519-signed update manifests gate the updater
• 330cbbd feat(parity): durable paper execution ledger + weekly paper-demo parity snapshot (observational)
• e9d6b7a refactor(lifecycle): P0 seam S10 — dispatch and protection leave the god module
• 92e0620 chore(docker): move redis behind explicit profile — future-only infra, no default startup barrier (P2)
• 74dc9f8 feat(news): version sentiment pipeline provenance + deterministic high-impact tagging (P0 2A/5A)
• 601bbf9 fix(qa): portable mutation runner - repo-relative resolution, sys.executable fallback, --list-targets introspection
• a08a91a feat(execution): canonical execution-cost assumptions — one source of truth (CAL-2026-09-07-A)
• 63ff20e P1 seam L7: post-policy pipeline stages leave live_engine
• 96fb122 feat(research): streaming-replay candidate benchmark — causal counterfactual A/B/C substrate
• f2b4a0b feat(economics): canonical EconomicAssumptions + SizingPolicy (ECON v1)
• a5f3846 feat(paper): REPLAY mode — real historical chronology as an explicit, fail-closed paper data source (P0 phase 4)
• 09e2ffa feat(calendar): add forward-looking economic event layer (ECON_EVENT v1)
• 955971b feat(risk): profit-protection circuit breakers — daily/weekly loss budgets + consecutive-loss cooldown gate NEW entries only
• cfcb8cb feat(risk): calibrated confidence before risk scaling — raw softmax no longer levers money-at-risk (P0 phase 3)
• 10d8e25 style(live+execution+indicators): ruff format conformance on the L-wave/P0-wave extraction set
• 951dcf5 fix(lint): import-order repair in model contract migration test (P0-wave residue)
• 75b7c69 fix(lifecycle): ChampionModel/ChampionManager default head width follows the canonical 3-class contract (P0 phase 2 follow-through)
• 7427219 fix(live): import torch/numpy in inference service, drop unused contextlib
• 7f51116 fix(live): import Sequence in inference service (L6 extraction residue)
• 891f78d perf(research): BUG-106 PHASE-2 — cursor-incremental pool evidence + prefix-stable HTF aggregation in the 70D builder
• 3edb9bb P1 seam L6: inference service leaves live_engine
• ddf736a refactor(models): canonical 3-class head is the ScalpNet default — dead WAIT logit now opt-in legacy compat (P0 phase 2)
• 664e433 fix(lint): sort imports in schema_70d_reconciliation test (L1 contract conversion residue)
• 5ef93f9 fix(lint): drop unused RegimeState import + format live_engine (L-wave residue)
• 5c80e55 fix(lint): un-sort policy threshold-ownership test imports (P0 wave residue)
• 1ac3d61 P1 seam L5: completed-bar processing leaves live_engine
• 94b8753 fix(lint): drop unused deque import in order_manager (L-wave residue)
• 93ed951 fix(live): repair L3/L4 extraction imports — hot_swap + maintenance missing names
• d57d003 fix(policy): single-source confidence thresholds — remove divergent 0.20/0.95/0.70 literal defaults (P0 phase 1)
• dfe4e21 P1 seam L4 (recovery): re-apply hot-swap delegate after concurrent live_engine overwrite
• 3cf05dd P1 seam L4: hot-swap service leaves live_engine
• 2c26e7d fix(gate): WinError-206 argv guard in check_local stages
• 7b24f27 fix(lint): drop TradeOrder import left unused by the L2 extraction
• becd73e P1 seam L3: periodic maintenance cycle leaves live_engine (dedupe x2 call sites)
• 0fc9531 P1 seam L2: decision-execution stage leaves live_engine (tick pipeline tail)
• 97cde71 refactor(gate): beforePush.sh becomes a thin delegating wrapper
• e307e0c test(qa-deep): flip missed BUG-194 crash probes to the post-fix contract
• 1566bf9 fix(db): fail-loud migration path + migration-safety CI check
• 8943b31 ci(deplock+qa): drift + migration-safety lane; weekly mutation gate fail-loud
• 7466ac3 test(mutation): honest mutation runner + dependency SSOT (uv lock + generated artifacts)
• 9a1f3d8 refactor(live): L1 shadow recorder extraction — wiring repairs + contract test conversion
• 5ca201c P1 seam L1: shadow recording extraction recovered + verified
• 90a6c0b fix(lifecycle): repair S9 scoring extraction wiring — missing trajectory-features delegate, N805 staticmethod, lint/import hygiene
• c00ed65 refactor(update): split updater responsibilities into release/update_engine package
• f51a837 refactor(lifecycle): P0 seam S8 — reconciliation + autopsy + experience leave the god module
• 7fab446 refactor(lifecycle): P0 seam S7 — pending-order lifecycle leaves the god module
• 8337419 types(execution): strict-clean execution package — honest S5 view/ledger accessor types (mypy --strict: 89 -> 0 errors); golden accessor test strengthened to live-identity invariant
• cbab360 refactor(lifecycle): canonical TicketState view layer + single-default presence semantics
• f439349 refactor(lifecycle): introduce TicketStateStore (P0 seam S5)
• a5ddc75 fix(audit-writer): drain-then-wait batch loop — same 500-batch semantics, 1 idle Empty per interval
• e178d2a fix(workers): 5s idle waits for candle-intel + shadow70 — debugger walls cut, dispatch immediate
• bee4942 fix(telegram): 3s blocking idle wait — 15x fewer first-chance Empty walls, dispatch untouched
• c1f4eff fix(db): repo-wide PRAGMA-gated ADD COLUMN — factory/research stores join the zero-exception pattern
• 0977a89 fix(db): fleet-wide PRAGMA-gated ADD COLUMN — ends duplicate-column walls for good
• e491559 fix(db): PRAGMA pre-check skips existing columns — zero duplicate-column exceptions
• 7874bcb fix(release): pre-flight shutil.which guard + import sort — zero-spawn wmic path, debugger-clean boot
• fd84327 fix(release): _run swallows FileNotFoundError — missing wmic no longer kills startup
• aca798c fix(lagstorm): reseed-guard + vectorized liquidity lifecycle + telegram 401 breaker
• caf9f3a fix(indicators+web): end 20k lag storm — streaming StochRSI O(n) (2.6s->0.2s), non-overlapping 30s poll, 15s abort, 10s error cadence
• dd73524 feat(indicators): 20000-candle window — O(n) MACD rewrite (27s->1s), API limit 20k, engine seed 20k, widget fetches 20k
• f8d38b0 fix(indicators): TV-parity vote layer — directional MOM/BBP/W%R, K-vs-D Stoch/StochRSI, W%R abs display, ICH deadband, classic/woodie R3=R2+range
• 8482ab3 docs(indicators): register agents/xauusd_indicator_math.md as Section A-D math spec
• f936406 test(indicators): xauusd_indicator_math.test.js — 31 API regression tests (11/15/7 shape, Section-D parity, Williams canonical/legacy, DM gaps, all 10 timeframes)
• 023b23c feat(indicators): service wires xauusd_spec across all timeframes — H/L/Vol threads, TV vote maps, _sign_vote, Section-D gauges (rating +-0.5/+-0.1)
• 5b09ad0 feat(indicators): xauusd_spec.py — H/L-aware Section A-D helpers (ADX Wilder DI, AO rise/fall, MACD-vs-signal, W%R canonical, BBP Elder, UO, VWMA, HMA, Ichimoku, Section-D buckets)
• 74023c7 fix(indicators): RSI>70 Sell / <30 Buy + CCI>100 Sell / <-100 Buy per xauusd spec A1/A3 (was inverted)
• 5c2a442 feat(web): add TV-style triple cycles on top — additive to distribution bars
• d1686e5 e2e(smoke): full upgrade of the end-to-end smoke chain to whole project
• 7eaf823 fix(web): serve tv_widget_styles.css — missing FileResponse route caused 404 and unstyled card
• b90e5db test(web)+fix(web): TV-REDESIGN-1 JS contract suite (16 tests) + tfchip context chip; comment hygiene
• ae60f89 Hermes-Main: rebuild tailwind.css after TV-REDESIGN-1 markup swap (canonical build script, purge audit clean)
• ffeff26 feat(web): TV-REDESIGN-1 professional terminal card — market-context strip, distribution-bar signals, badge tables, real stale/error states
• 0ac603b Hermes-Main: register TV-REDESIGN-1 change entry (UI-only scope, gauges->distribution bars approved)
• 8710ce4 fix(web): remove realtime candle values block from chart panel
• a930ff6 feat(paper-reality-p2): Priority 2 durable recovery — persist balance/positions/tickets
• f10e67e fix(web): unbreak page scroll — overflow-x hidden on html+body was the lock
• 0af1fa5 style(web): sleek modern global scrollbar
• 4d6e9a4 feat(paper-reality-p2): Phase 2 — auto SL/TP + calibration + execution proof
• bd9f4bb feat(paper): deterministic stress market + real fills (BUY@ask/SELL@bid, slippage) + live accounting
• 9391350 feat(indicators+web): native-dark technicals card + real multi-timeframe engine
• 90af214 feat(indicators+web): TradingView-exact M1 math + card UX pass + window scroller
• 3142942 fix(web): fix locked scroll + card contrast + active pill
• 8ba8563 fix(web): restore white-card text + unlock page scroll
• f810314 fix(web): inline TradingView indicator widget + serve tv_widget assets (no 404)
• 49699c4 feat(indicators): SOLID indicator engine + live homepage widget (TradingView-exact)
• 382b616 fix(gateway): ruff lint/format + gateway E2E suite (HMAC, 14 actions, CLI, daemon)
• 9e69b4e feat(gateway): add Windows MT5 bridge server for existing Linux gateway client (Demo, HMAC contract)
• 74e5b32 Nexus-Main: fix walk_forward NaN-buffer contract — test now expects fail-closed ValueError (BUG-243B)
• 060eec8 Nexus-Main: smoke CI hardening — mypy + determinism fixes (post-CI 33994376976)
• 7c14094 Nexus-Main: production-grade E2E smoke and runtime verification hardening (CHG-SMOKE-E2E-V2)
• cd5544e Nexus-Main: forensic closure ledger fix — HEAD 317a82e, v2 smoke stash, commit chain 3527442→14d5270e→317a82ec
• 317a82e Nexus-Main: forensic stash reconciliation closure — 91→2 prunes, 91/91 refs, 13/13 tests, P0 fixes verified
• 14d5270 Nexus-Main: forensic p3 archival — 5 untracked trees inventoried, safe-to-prune gate (45MB tars excluded, reachable via forensic refs)
• 3527442 Nexus-Main: forensic stash reconciliation — 85-stash audit, 91/91 backups, Phase 7 request (READ-ONLY)
• 59c7814 Nexus-Data: ruff format fix — walk_forward_trainer.py (Agent-8 BUG-243B cherry-pick residue, CI-parity)
• d879254 Agent-8: BUG-243B trainer fail-closed hygiene — non-finite training frames refused (ecosystem-clean W2)
• 7a07491 Agent-8: BUG-243 retrain-buffer mixed-width None-poisoning guard + regression suite
• 8a6ff51 Nexus-Data: BUG-243/243B RED baseline — Agent-8 regression suites (7 tests) staged pre-fix, RED on unpatched src
• 3fcc150 Hermes-TDFQ2: NX-TDFQ2-HERMETIC — de-machine-state live.yaml freshness assertion (tmp_path synthetics)
• 865e3b5 Hermes-EngineGuard: NX-GATEPARITY-PARITYFIX — local-only forensic note + xdist flag alignment
• d6e82c3 Nexus-Main: CI 941 repair — mypy list[str] annotation + restore a2-lineage CSV skip guard
• 1b5d4a7 Nexus-Main: CI repair on 141b7f8 — B017 narrowed + ruff format on 6 Wave-2 files
• 141b7f8 Wave-2 registries: taskboard row TASK-W2-ECO-GATING + CHG-0061-W2-GATE ledger closure
• b7dd77a Wave-2 ecosystem gating: critical_suite pins the fetch->clean->train + dataset hardening contracts; absorb subagent forensic + QA verdict docs
• f84a050 Nexus-Main: fix a2-lineage bounded test rerun — real artifact cleanup instead of missing delete_dataset
• 5be79e5 Agent 16 W2: runtime bars hygiene — naive-ISO fallback + double-Z strip + 13-test ecosystem regression (CHG-0064-W2)
• 5a2e553 Nexus-Main: lint repair on test_agent14 — B017 narrow + exception taxonomy imports
• 105229c Agent-3 Wave-2: ruff format + mypy heal for LEARNFIX-3 builder wiring (schema_v2 + incremental)
• e6d7c35 Agent-3 Wave-2: LEARNFIX-3 wiring — normalize_bars_frame at all four builder entries (bars_frame_to_bardata + compute_60d + compute_70d canonical + compute_70d_frame_fast); restores the runtime-cleaning contract lost in a parallel branch hop
• 0699bd8 Nexus-Main: CI-only guard — skip a2-lineage bounded test when raw CSV absent
• 654c5c3 Nexus-Main: lint fix — B017/F841/RUF100/I001 on 3 CI-flagged files (CI 938 red)
• 3941ebc Nexus-Main: ruff_format+lint fix — last 2 CI-flagged files (CI 936 red)
• 89e7f43 Nexus-Main: ruff_format fix — 10 remaining CI-flagged files (CI 934 red)
• 805875f Agent-3 Wave-2: LEARNFIX-3 — runtime bars normalization for multi-user fetch (bars_normalize + builders wired; CSV strings/epoch ms+us/naive datetimes/dup/unsorted/NaN all cleaned at runtime; honest failure, never fabricated bars; shape-transparent feature frames; fast/canonical byte-identity preserved; 6-test regression suite)
• 551a252 Nexus-Main: TASK-SMOKE-E2E taskboard row (VERIFIED at origin/main)
• 3b049ee Nexus-Main: CHG-0065 registry close-out + taskboard row (SMOKE-E2E verified at origin/main)
• e0a489e Agent-1 Master-Model-Forensics: fix mypy attr-defined in StreamingReplayEngine pending provenance (created_at: object -> datetime)
• 18899f3 Nexus-Main: ruff format+lint fix — 18 CI-flagged files (CI 928/929 red)
• 75ce90e Nexus-Main: wire E2E smoke chain into CI quality gate + critical-suite manifest
• 6a9c8b5 Agent 12 Wave-2: BUG-250 code fix - reason==0 profit-alone no longer MANUAL_CLOSE
• 5c34825 Agent 12 Wave-2: BUG-250 exit-classification INV-012 hardening (reason==0 profit-alone no longer MANUAL_CLOSE)
• 17d43ad Agent-4: BUG-243 regression suite (4 tests: canonical-3 / legacy-4-only-when-declared / missing-meta fallback / garbage fallback)
• 5bbea56 Agent-10: BUG-243 model-pipeline hardening — end-to-end persist refusal + stale-retrain discard + hot-swap meta/tensor coherence gate
• e3d6d57 Agent-1 Master-Model-Forensics: PHASE-7 replay A/B — champion(4-head) vs repaired pilot(3-head) on identical 600-bar window (evidence)
• 9edbfe7 Agent 6: BUG-249-A6 ledger row — execution clamp NaN/inf leak (wave 2 QA)
• 924a40a Agent-4: BUG-243 regression suite - bundle-coherent class-head mint (canonical-3, legacy-4-only-when-declared, missing-meta fallback, garbage fallback)
• 60b1f42 Agent 6: BUG-248 NaN/inf guard on OrderLifecycleManager._clamp_dispatch_volume (wave 2)
• 2a4cdaa Agent 2 (Nexus-Main): BUG-244 follow-up — trainer-exclusion test aligned to actual CandidateTrainer train-pool semantics (train pool = non-test, non-purged rows; purged excluded from BOTH pools)
• 531af1d Agent 15: land W2 strong-QA handoff on main (CHG-0059-W2)
• d1c5a2e Agent 17: taskboard W2 evidence row - leakage injection + governance sweep + 89-test scope gate (CHG-0063)
• 09d2952 Nexus-Main: promotion-gate evidence chain on c9982ddde175 — OOS tables + parity + determinism; Agent-1 governed promotion verified
• 19e786d Agent 15: taskboard close — TASK-AGENT15-BTREPLAY-W2 VERIFIED (CHG-0059-W2)
• 9c302c2 Agent 15: W2 QA — fix test_agent14 import + exception taxonomy (CHG-0059-W2)
• 9bc9af2 Agent 17: ruff B905 zip strict + format on test_agent17_oos_hard_gate (W2 sweep)
• 0889f25 Agent 17: ruff B905 zip strict + format on test_agent17_oos_hard_gate (W2 sweep)
• 8fe4fdf Agent-4: BUG-243 bundle-coherent class-head mint — declared head replaces hardcoded 4 in live_engine (3 mint sites + helper) + shadow/model_governance inference + debug routes
• fbadcf9 Agent 13: bugs.md BUG-249 ledger row — research methodology fail-open trio (append-only; id-collision note vs BUG-244 bar-mode)
• 52d874f Agent-4: flip BUG-194 battery to the post-fix contract (bfeb08d landed)
• 2e3a762 Agent 15: register TASK-AGENT15-BTREPLAY-W2 + CHG-0059-W2 (strong QA wave 2)
• 69033c5 Agent 2 (Nexus-Main): dataset split-boundary horizon purge — BUG-244 (train/val/test boundary no longer split-adjacent)
• b2049f4 Agent 18: taskboard — TASK-AGENT18-REPLAY IN_PROGRESS -> VERIFIED + handoff (CHG-0062, 3/3 GREEN)
• 01a94ae Agent-11: handoff doc — execution/risk forensic wave (CHG-0064)
• 4a46fdd Agent 13: BUG-244 fail-loud negative purge/embargo + zero-finite OOS fail-closed + provenance forwarding (research methodology mission)
• 8735c34 Agent-11: 60-scenario router coverage contract suite (6 tests)
• 1cb1c7c Agent 18: handoff — REPLAY ENGINE forensic (CHG-0062, LIMIT pending-queue, dataset==replay bit-exact)
• b6c75a4 Agent 12: BUG-247 code fix - hedge clamp + AI-flip SAFE_MODE gate (CHG-0058 follow-up)
• 1dd6042 Agent-11: land CHG-0064 change-control entry + 14-test regression suite
• 7c821a4 Agent 2: fix BUG-246 — dataset builder gap SSoT bind (regen_70d_clean_dataset 900s->600s SSoT import)
• 5fe745c Agent 12: BUG-247 code fix back on main (re-land after addendum absorption)
• fa93c76 Agent 12: addendum — BUG-247 residual hardening (CHG-0058-FOLLOWUP, CHG-0065 doc layer)
• 2446475 Agent 12: taskboard FIXUP row for BUG-247 (append-only)
• a1fb7c3 Agent 12: CHG-0058-FOLLOWUP change-control row for BUG-247 (append-only)
• bfeb08d Agent-4 forensic fix: BUG-245 - guard the candidate-confidence division in signals/policy.py (live ZeroDivisionError / negative-confidence path)
• f33c434 Agent 12: BUG-247 ledger row (append-only; execution residual gates row)
• 22091be Agent 14: ruff format on dataset hardening + test alignment (CHG-0061)
• fcff9f7 Agent 14: bug ledger — BUG-239-DATASET audit (replay-cache dataset identity class, CHG-0061)
• 89bbcd5 Agent 14: dataset test-suite alignment — availability semantics + taxonomy pins (CHG-0061)
• b1eb5b9 Agent 5: research-gate hardening — fold OOS floor + SSOT purge defaults + smoke-scoped emission gate (BUG-244)
• a5ba9ca Agent 14: test contract alignment — empty-window availability semantics (CHG-0061 follow-up)
• 7e06278 Agent 18: streaming replay LIMIT pending-queue — green-after RED 18A/18C
• 4cba9fa Agent 6: BUG-239 + BUG-240 web control-surface fixes (minimal 4-line patch)
• 8ff61ff Agent 14: dataset acquisition/immutability hardening — MT5_TICK_DATASET v3 (CHG-0061)
• d53660e Agent 16: add handoff doc — walk-forward / purge / embargo / leakage forensic (CHG-0064)
• caf1c2c Agent 6: RemoteMT5GatewayAdapter dispatch parity — implement missing IMT5Port order contracts
• 27d7a33 Agent 15: BUG-244 regression suite — bar-mode SELL TP containment (4 tests)
• 7070c74 Agent 17: BUG-245 ledger row + OOS-gate mission handoff + taskboard completion (CHG-0063)
• c783cc1 Agent-11: regression suite 14 tests + bugs.md CHG-0064 cross-ref row
• 32194c9 Agent 16: register CHG-0064 + TASK-AGENT16-WALKFORWARD (walk-forward/purge/embargo/leakage mission)
• 4549847 Agent 6: BUG-239 + BUG-240 web control-surface fixes (blob-level commit defeating restore war)
• d8ed39a Agent 17: ruff format on test_agent17_oos_hard_gate (CHG-0063 follow-up)
• 977e8ad Agent 16: fix pipeline purge/embargo provenance forwarding + fabricated-benchmark-fallback removal (CHG-0064)
• 3f5bef2 Agent 15: fix BUG-244 bar-mode SELL TP containment — >= ev.low -> <= ev.high (P0)
• 390e153 Agent 17: fix OOSGate empty-context silent-widening + hard-gate regression (CHG-0063)
• 1f3aa50 Agent 7: handoff doc — AGENT 7 DATA-FLOW forensic (CHG-0060, TASK-AGENT7-DATAFLOW; TDF-1..9 + TDF-F1 evidence and verdicts)
• f355c54 Agent 12: execution-pipeline forensic regression net — 18-test invariant battery (CHG-0058)
• 13aeebd Agent 7: taskboard row — TASK-AGENT7-DATAFLOW mission COMPLETE (TDF-1..9 evidence probes, TDF-F1 test fix, boundary verdicts; CHG-0060)
• 88c8c32 Agent-4: identity-probe evidence — live champion c8c0b5b0 vs trainer/pilot bundles (scaler b3c65b65 shared pilot-live; t70d meta provenance-poor; 3/4-class split documented)
• b50b854 Agent-1 Master-Model-Forensics: fix replay 3/4-class integration blocker — infer num_classes from checkpoint tensor (P0 #3)
• f826db6 Agent 18: CHG-0062 stage — change-control append (pending-queue fix note)
• c07fc6e Agent-3: LEARNFIX-2 — honor WalkForwardTrainer.focal_gamma constructor arg (was silently hard-coded 1.0, declared default 2.0 discarded)
• e48803f Agent 18: RED regression suite — replay LIMIT execution fidelity (pending queue missing)
• 6599e4d Agent 7: fix TDF-F1 env-dependent regime-freshness test — live.yaml is git-ignored operator config; pin default+base.yaml+conditional override instead of operator file shape (red-before/green-after on machine with minimal live.yaml)
• 372d446 Agent 17: register CHG-0063 (OOS gate / temporal holdout / forward-test forensic change entry)
• ad17953 Agent 17: taskboard row - TASK-AGENT17-OOS-GATE (OOS/temporal-holdout/forward-test forensic mission, CRITICAL)
• d0a9b6d Agent-11: fix BUG-242 operator close/modify bypass — web routes through OrderLifecycleManager (INV-004)
• 1f5af30 Agent 18: register CHG-0062 (replay-engine forensic mission change entry)
• d6d398d Hermes-QA: taskboard row - TASK-NX-STP0-QA (State-Truth wave V1-V4 validated, verdicts PASS)
• ad06738 Agent-11: fix BUG-240 engine-wide exposure gate + BUG-241 SAFE_MODE dead on primary dispatch path
• 7715e4a Agent 14: RED regression suite — dataset acquisition/immutability contracts (CHG-0061, 19 tests red-before)
• 1a355b5 Agent-3: register TASK-AGENT3-MODEL-GOV (model governance / champion / 50D-70D contract forensic mission, CRITICAL)
• 0b6668e Agent 7: TDF-6..9 — real end-to-end trace (tick->50D->scaler->model->regime->policy->risk), risk-boundary sizing + clamp evidence, confidence-gate boundary walk, 14-injection contract-breaking matrix (all boundaries held)
• 152e8eb Agent-11: fix BUG-239 risk_engine.evaluate_proposal UnboundLocalError + micro-account margin hole
• 541c43c Agent-3: LEARNFIX-1 regression suite — provenance binding precedence + honest unbound + gate-unchanged pins (6 tests)
• 56a6843 Agent-3: LEARNFIX-1 — bind dataset provenance in train_and_validate (P0 residual: emission gate aborted after full training, artifact discarded)
• d95524d Nexus-Main: fix BUG-238 — shadow promotion_verdict NameError on noise-band branch (lazy-import alias missed one return)
• 710af35 Agent 18: register TASK-AGENT18-REPLAY (replay-engine deep forensic mission, CRITICAL)
• 8c73d7d Nexus-Main: wave-2 executed — full-history 34x10 isolated retrain PASS (producer c9982ddde175, champion c8c0b5b0 untouched)
• 46380b8 Agent 7: TDF-1..5 real-pipeline probes — 50D/70D contract, validator reachability, model+confidence semantics, hot-path timing, order-authority isolation scan (all PASS; evidence probes under scratch/)
• ebc07d9 Agent 14: register CHG-0061 + TASK-AGENT14-DATASET (dataset subsystem deep forensic mission)
• b635a36 Agent 15: register TASK-AGENT15-BTREPLAY + CHG-0059 (backtest/replay deep forensic mission)
• 2d03066 Agent 12: register CHG-0058 + TASK-AGENT12-EXEC-FORENSIC (execution pipeline deep forensic mission)
• 30a4f44 Nexus-Main: taskboard row — TASK-AGENT9-70D-RESEARCH claimed (full 70D research/validation/readiness mission, CRITICAL)
• 54d1e90 Nexus-Main: taskboard row — TASK-CODEQL-ZERO (926 -> 0 open alerts, campaign summary)
• d9580b9 Hermes-StateTruth: taskboard row — NX-STP0-CODER S1-S3 wave complete
• 04523d9 Hermes-StateTruth: S3 — forensics _make_engine hermetic (tmp_path artifact under artifacts/model_generation/models, CWD-independent yaml)
• b8d225a Hermes-StateTruth: S2 — failure-injection suite for taxonomy UNKNOWN vs verified-absent NOT_CONFIGURED
• 75a91a3 test(state_truth): relax cross-context dimension parity to resolver honesty
• 4dd2bbf Nexus-Main: fix test_matrix_model_fields_match_snapshot fixture-scoping flake
• 842b3d1 Nexus-Main: CodeQL batch 5 — build_site format cleanup + matrix test doc comment
• fd386dd security(sql-guard): block-comment rule + DDL verb allow-list; state_truth unavailable passthrough
• 4760f33 style(selftest): paren-free iteration in _safe_assert_text (ruff RUF-style)
• 628c274 Nexus-Main: CodeQL batch 4 — selftest safe-assert taint-laundering rework
• 03899bc Nexus-Main: CodeQL batch 2 — repeated-import removal + parallel batch-3 validation commit
• 0002927 security(docs): sanitise release cache at load boundary — fixes CodeQL #119 properly
• 9e0f418 Hermes-StateTruth: S1 — TestCommitIdentity pinned to MAIN RULING (repo-root stamp stale in dev; foreign-location stamp WINS; dirty-tree mirror cases)
• 0bed1da Hermes-StateTruth: S1b — parity test version pin 9.0.6 -> pyproject-derived canonical (R4)
• 41642fd Nexus-Main: empty-except sweep batch 1 — contextlib.suppress (72 files)
• 67f34c0 fix(marketplace): drop shadowing local import contextlib (ruff F823)
• 1fec286 security(codeql): sanitize remaining stack-trace sinks #121-#123
• 209af8a security(codeql): selftest AssertionError keeps synthetic diagnostics
• 37ae4d3 security(codeql): fix ERROR batch — taint, cycles, exports, tarslip
• f491832 security(codeql): restore selftest assertion messages for guardrail
• c6ee3a3 security(codeql): harden db_console remainder — stack-trace + schema bound param
• c589966 security(codeql): harden db_console surface + stack-trace exposure
• b7b422f security(codeql): fix py/sql-injection #108 — db_console table handling
• 00ae8a0 security(codeql): fix py/stack-trace-exposure #124 via sanitized validation
• 3db0ef5 security(codeql): suppress false positive py/clear-text-storage on docs builder
• f00265c security(codeql): fix py/stack-trace-exposure #125 via generic error + server log
• 80c3e8f ci(security): run CodeQL on pushes to main
• 96735e9 security(sqlite_driver): fix CodeQL py/sql-injection on table_columns
• 7228849 Add MIT License to the project
• 854c894 docs(readme): enrich top badge bar — big clickable star + commits/size/forks/issues
• 069f760 docs(readme): move star ask + Star History higher
• 681b08a docs(readme): irresistible star ask + polished Star History chart
• 366cf33 docs(readme): add pretty star ask + Star History chart
• f3e98fe docs(readme): fix non-clickable Start here — anchors around text nodes only

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.