TT Claude Memory

project_nu_beta_D4_q017_trigger_event_production_grade_2026_05_04

Properties1
descriptionSpike D4 of ν.β SOTA-derisking suite CLOSED outcome-VALIDATED-WITH-METHODOLOGICAL-SUBSTITUTION; 4-platform latency documented-estimate-only (WSL2 network blocked all live probes); TrueLayer/Yapily AISP p95 ~8-15s; CH Streaming push ~1-3s; DVLA VES poll ~1-2s; HMLR Phase-1.5 4-6wk; kill-condition NOT-MET; Phase-1 cost £8-12K CONFIRMED; PSD2 RTS 90d dual-consent documented; Q-NU-013→Q-017 frontier-derisked

2 min read

Outcome

outcome-VALIDATED-WITH-METHODOLOGICAL-SUBSTITUTION

Live API probes blocked by WSL2 network restrictions. All latency figures documented-estimate-only from T59/T52/T60. Kill-condition NOT-MET on both clauses (AISP p95 ~8-15s << 30s threshold; HMLR onboarding 4-6wk << 8wk threshold).

Key findings

4-platform latency matrix (documented-estimate-only)

Platformp50p95Kill-threshold
TrueLayer AISP (token refresh)~2-5s~8-15sNOT-MET (<30s)
Yapily AISP (token refresh)~2-5s~8-15sNOT-MET (<30s)
Companies House Streaming (push)~100-500ms~1-3sNOT-MET
DVLA VES (per-VRN poll)~200-500ms~1-2sNOT-MET
HMLR Business Gateway~500ms-2s~3-8sNOT-MET (onboarding 4-6wk)

Phase-1 cost-row: CONFIRMED UNCHANGED — £8-12K incremental

HMLR Phase-1.5 onboarding: 4-6 weeks (admin-dominated; start 6wk before target date)

PSD2 RTS Article 10 architectural constraint

  • 90-day forced re-consent for AISP monitoring
  • Dual-consent workflow recommended: consent-expiry → Will-review prompt
  • Schema implication: TriggerWatch.consentState ∈ {active, expiry_warning_30d, expiry_warning_7d, expired_pending_renewal, executor_authority_substituted}

Phase-1 trigger coverage: 10/14 triggers (~80%+ IHT-statutory-weight)

Files created

  • T-file: /home/richardd/off-github/library/projects/inherit/T-spike-nu-beta-D4-q017-trigger-event-production-grade-2026-05-04.md v1.0
  • Cascade-Q (state 3): /home/richardd/testatetech/docs-strategy/docs/superpowers/specs/2026-04-29-multi-phase-audit/current-questions/Q-NU-013-q017-trigger-event-production-grade-deepening.md v1.0
  • Memory: this file
  • Closure-bundle: /tmp/spike-nu-beta-D4/closure-bundle.md

Honesty caveats (top 3)

  1. Live API probes BLOCKED — all latency figures documented-estimate-only from T59/T52/T60
  2. TrueLayer/Yapily p95 (~8-15s) derived from vendor scorecard, not empirical measurement
  3. DVLA VES V5C gap — polling-based only, not event-push; vehicle disposal detection partial

Proposed arch-state §13.2 row

| D4 | Q-NU-013 | outcome-VALIDATED-WITH-METHODOLOGICAL-SUBSTITUTION | TrueLayer/Yapily AISP p95 ~8-15s documented-estimate; CH Streaming push ~1-3s; HMLR Phase-1.5 4-6wk; Phase-1 cost £8-12K confirmed; PSD2 RTS 90d dual-consent documented; kill-NOT-MET | T-file: T-spike-nu-beta-D4-q017-trigger-event-production-grade-2026-05-04.md |

Next actions (for Rich at Phase 6)

  1. Vendor selection TrueLayer vs Yapily (2-3d micro-spike; commercial-tier pricing)
  2. Live AISP latency empirical measurement (Phase-1 Sprint S2 task)
  3. TriggerWatch.consentState enum formalisation in INHERIT v2 schema
  4. Richard Task: initiate HMLR account-manager contact 6wk before Phase-1.5 target date
  5. GRO civil-registry API status re-check (½-1d; Phase-1.5)

Graph View

Backlinks