R-Dash v1.2.0 verdict = PILOT-BLOCK. Execute 3-wave remediation (W13 TRUTH-UP wk1-2 → W14 SECURITY HARDENING wk3 → W15 UX POLISH wk4-6) before any deployment to function heads. Tag v1.4.0 as the GA ta
Decision
R-Dash v1.2.0 verdict = PILOT-BLOCK. Execute 3-wave remediation (W13 TRUTH-UP wk1-2 → W14 SECURITY HARDENING wk3 → W15 UX POLISH wk4-6) before any deployment to function heads. Tag v1.4.0 as the GA target. Adopt 5 premortem-derived mitigations: parallelize W13 doc-fixes + Snowflake-worker + error-rename tracks; ship admin-managed claims-mapping as RLS AD-integration fallback; book external pen-test on W13 day-1; add Cube fork-or-replace scoping spike to W14; capture p95 widget-data + first-task-time baselines on W13 day-1 before any changes ship.
Rationale
Cross-agent consensus from 5 independently-contexted sub-agents converged on 12 CRITICAL findings with file:line citations (highest signal possible from a non-deployed audit). Bias scan: 0 patterns. Counter-argument considered (steelman: ship pilot now to gather signal) and rejected — 3 risks are reputation-irreversible: (a) LICENSE assigns Runwal copyright on AJ IP and cannot be corrected post-pilot without admitting misclassification; (b) RLS theatricality + claims={} means a single demo-day mishap leaks salary data, unrecoverable; (c) home-page “under construction” + JSON health card burns the one-shot first-impression budget across 9 executive stakeholders. Premortem identified 5 risks across execution/assumption/people/external/design categories; each has a concrete mitigation that fits within the stated calendar windows. Remediation work is additive + reversible. The 4-6 week investment is small versus the cost of a pilot pulled mid-flight after a credibility-damaging incident.
Alternatives Rejected
Outcome
Pending