The Hub Architecture is a prime-structured compute field. The CAL Suite is the active runtime that routes work through that field, executes it, and reports not just the result but the path the field took to get there.

This repo is the working implementation of that system.

• Hub Architecture: a structured field of prime-based hubs with roles, phases, HMV classes, and routing behavior.
• CAL Suite: the software runtime around that field.
• Hubs: active compute nodes in the field, not passive labels.
• Routing: work is sent through the field according to structural fit, resonance, and fleet posture.
• Holonomy: CAL reads the journey of a computation or analysis path, not only its endpoint.

The core idea is simple: the architecture does not only compute on mathematical inputs. It computes through mathematical structure.

The CAL currently operates a 111-hub six-fleet field with pinned Pillar identity:

1. Origin Anchor (1): Universal reference point / Fixed point of emanation.
2. Foundation Fleet (22): Origin-scale primes (2-71) + dense anchors (137, 613).
3. Balanced Sovereign (10): High-level framing hubs at 10^9 scale.
4. Fleet Beta (22): Navigation, bridging, and mediation hubs at 10^9 scale.
5. Fleet Charlie (22): Tactical local resolution / compressed mid-field handling.
6. Fleet Delta (22): Boundary, fortification, and Mersenne-adjacent control.
7. Legacy Fleet (10): Historical HMV-5 pathway / deep-field route spine.

Total: 111 Hubs — The most structurally complete hub mesh in the archive.

• Pillar-based routing
• HubQL field inspection
• H-ISA program execution
• native holonomy and route analysis
• advanced topological lens mapping for deep multiplicative structures (e.g., Riemann Zeros)

For the current authoritative runtime posture, read CAL_SINGLE_SOURCE_OF_TRUTH.md.

• ComputeAbstractionLayer: main runtime entry point for routed computation, H-ISA execution, and probes
• CALWorkbench: higher-level reporting and structural analysis
• HubQL: query language for inspecting hubs, routes, and configurations
• H-ISA: the instruction surface for running structured programs through CAL
• holonomy: path, curvature, closure, drift, and journey-family analysis

Most of the implementation lives in cal.

from cal import ComputeAbstractionLayer, CALWorkbench, HubQL

cal = ComputeAbstractionLayer()
result = cal.gcd(84, 30)
print(result)

workbench = CALWorkbench()
report = workbench.number_theory_profile(360, agent_sv=5.0)
print(report.summary["average_resonance"])

hubql = HubQL(registry=cal.registry)
rows = hubql.execute(
    "SELECT hub, fleet, hmv, role FROM HUBS "
    "WHERE runtime_reachable = true ORDER BY runtime_hits DESC LIMIT 5"
).rows
print(rows)

python tools/cal_cli.py gcd 84 30
python tools/cal_cli.py miller-rabin 2147483647
python tools/cal_cli.py hubql "SELECT hub, fleet, hmv FROM HUBS WHERE runtime_reachable = true LIMIT 5"
python tools/cal_cli.py hubql "SELECT route_key, tri_sync, route_score FROM ROUTES"
python tools/cal_workbench.py nt-profile 360 --sv 5.0

• CAL_SINGLE_SOURCE_OF_TRUTH.md
• BASIN_CENTRE_CHARACTERISATION_2026-04-05.md
• HUBQL_GUIDE.md
• PILLAR_ROUTING_ARCHITECTURE_2026-03-15.md
• BEER_PROCESS_HUB_FIELD_MAPPING_2026-04-01.md
• docs/architecture
• docs/testing

• cal: runtime, routing, HubQL, workbench, probes
• tools: CLI entry points and runnable probes
• tests: validation and regression coverage
• docs/architecture: architecture and operating posture
• docs/testing: probe notes, validation reports, findings

This README is intentionally tight. It is the front door to the Hub Architecture and CAL Suite, not a full research index.

If you want the deeper theory, probes, research notes, and domain work, start from docs.