leeway-quantum-readiness

Makes LeeWay skills quantum-aware and hybrid-ready without falsely claiming classical agents are quantum computers. Governs quantum/hybrid capability detection, backend provenance, circuit/job evidenc

SKILL.md

GitHub

LeeWay Quantum Readiness

Core law

LeeWay must be ready to use quantum computing when real quantum or verified quantum-simulation capability is available, while remaining fully correct on classical hardware.

Quantum-aware does not mean quantum-executed.

Never claim quantum speedup, quantum state, entanglement, amplitude behavior, annealing advantage, error correction, hardware execution, or quantum supremacy/advantage unless the relevant runtime and evidence actually support the claim.

Hybrid-first architecture

Preferred design:

LeeWay Context/Formula → capability classifier → classical path | quantum/hybrid path → runtime → Veritas → receipt

Every quantum-capable workflow must define a classical fallback unless the task is inherently quantum-specific.

Capability states

Use precise states:

  • QUANTUM_NOT_RELEVANT
  • QUANTUM_AWARE_CLASSICAL_ONLY
  • QUANTUM_SIMULATOR_AVAILABLE
  • QUANTUM_HARDWARE_AVAILABLE
  • QUANTUM_HYBRID_EXECUTED
  • QUANTUM_EXECUTED
  • QUANTUM_BLOCKED
  • QUANTUM_FAILED

Do not substitute simulator execution for hardware execution.

Quantum evidence contract

When quantum execution occurs, preserve when available:

  • provider/runtime identity;
  • backend/device identity;
  • SDK/compiler/runtime version;
  • circuit/program representation identity;
  • circuit/program hash;
  • transpilation/compilation settings;
  • qubit count / logical problem size;
  • shot count or execution sampling policy;
  • seed when applicable;
  • queue/job/execution ID;
  • calibration/noise snapshot identity when exposed;
  • mitigation method if used;
  • raw measurement/result artifact;
  • classical preprocessing/postprocessing;
  • comparison baseline;
  • Veritas measurements;
  • final receipt.

Math boundary

Keep these distinct:

  • theoretical quantum algorithm complexity;
  • simulated performance;
  • noisy-hardware observed performance;
  • fault-tolerant projections;
  • end-to-end application performance.

Do not turn asymptotic theoretical advantage into a real-world speed claim without measurement.

Error/noise law

Distinguish:

  • noise characterization;
  • error mitigation;
  • error detection;
  • quantum error correction;
  • fault tolerance.

These are not interchangeable.

Skill graph integration

A skill may declare relationships such as:

  • quantum_accelerable
  • quantum_native
  • hybrid_preferred
  • classical_fallback
  • post_quantum_security_relevant

Quantum capability is selected only when Context Engineering and Formula Governance determine that it materially improves the task and the runtime evidence supports the route.

Quantum-ready data contracts

Where practical, keep problem representations backend-neutral so future quantum adapters can consume them without rewriting the entire skill.

Separate:

problem definition → mathematical representation → backend adapter → execution artifact → verification

Avoid binding domain logic directly to one provider SDK unless the skill is provider-specific.

Post-quantum awareness

Quantum readiness also includes security transition awareness. When cryptographic longevity matters, distinguish current cryptography, migration planning, and post-quantum cryptographic requirements. Do not invent compliance requirements or algorithm status; consult current authoritative standards when making implementation decisions.

Emerging technology law

Quantum technologies evolve quickly. Treat provider claims, benchmarks, roadmaps and research announcements as discovery evidence until independently verified for the actual task/runtime.

Route new quantum capabilities through leeway-skill-lifecycle-governance before promotion.

Output law

When quantum capability materially affects a decision, explain it in plain engineering language:

what problem class → why quantum/hybrid might help → what backend actually ran → what was measured → what remains classical → what is still unproven.