Done. Report written to `tools/data/reports/agent_20260430_1946.md`.

**Risultante**: Primes are not weak GUE. The dipolar angle reference frame reveals three distinct regimes:
- **GUE/GOE**: theta = -97 deg, anticorrelation-dominated (dL1/dSR = 8.4)
- **Primes**: theta = -111 deg, balanced ordering (dL1/dSR = 2.28), 14 deg from GUE at z = 170
- **Poisson/Cramer**: no coherent direction (noise)

The prime ordering signal has a different *quality* than GUE — 3.7x more gap-similarity depression per unit of anticorrelation. The angle is locked at -111 +/- 1 deg across 3 decades of scale while the magnitude decays as ln(p)^{-0.14}. Previous report's -150 deg corrected explicitly (L3 lens).

**Consecutio**: can the ratio dL1/dSR = 2.28 be derived from Hardy-Littlewood pair correlations? If yes, it's a consequence of the arithmetic. If no, it's a new constraint on prime gap models.