What problem does it solve? AI measurement pipelines often accumulate multiple code paths that write to the same tables, making it impossible to tell which route created a row or whether a claimed pipeline phase actually executed. This Skill audits measurement runs to enforce a single canonical runtime path and requires full provenance stamping on every canonical row. ## Core Features & Use Cases - Canonical path enforcement: Verifies every run follows PullMeasurementsButton → useMeasurementJob.startJob → start-ai-measurement → autonomous-graph-solver → MeasurementReportDialog/render-measurement-pdf. - Provenance field validation: Checks that canonical rows carry created_by_function, created_by_component, solver_entrypoint, canonical_measurement_route, route_audit_version, and geometry_report_json.route_provenance. - Legacy route flagging: Blocks legacy routes from silently creating canonical-looking rows and requires route_warning = legacy_noncanonical_measurement_path. - Phase-execution honesty: Refuses to mark phases (3A, 3C, 3D, 3E) implemented without executed = true diagnostics or an explicit skipped_reason. - Use Case: Before merging a change to the measurement pipeline, run this audit to produce active and legacy route tables, canonical row proof, and a minimal migration plan. ## Quick Start Audit the current AI measurement pipeline and produce the active route table, legacy route table, and canonical row provenance proof before proposing any fixes.
