The complete reference, in one place.
Open the searchable protocol reader for every publication, core, consensus, execution, service, and compatibility document.
Open full protocol↗
PROTOCOL REFERENCE
Read full protocol ↗
Back to site ↗
NULS / KNOW THE GROUND BEFORE YOU BUILD
This reference introduces NULS as a protocol foundation and maps the current NULS Mars implementation workspace. It is organized around boundaries, choices, and responsibilities — not feature marketing.
HOW TO READ THIS REFERENCE
Start with the complete map, then follow the dependencies: shared principles, canonical data, network agreement, execution boundaries, and service modules.
Open the searchable protocol reader for every publication, core, consensus, execution, service, and compatibility document.
Open full protocol↗Understand what a network must preserve before choosing how it should evolve.
Read design principles↓Move from architecture to frontier modules and ChainBox without skipping the boundaries between them.
Follow the architecture↓DESIGN PRINCIPLES
NULS treats a network as an agreement with two kinds of content: the truths it must preserve, and the choices its participants must be free to make.
Protocol invariants, canonical serialization, and validity rules give every implementation a shared ground.
Consensus engines, VM capabilities, application modules, economics, and deployment targets should meet at explicit interfaces.
A network specification must be validated, versioned, and traceable — not scattered across secret scripts and undocumented defaults.
SYSTEM ARCHITECTURE
The current architecture describes a practical implementation map. The protocol remains larger than any one runtime or language.
SDK · CLI · wallet · explorer · AI service · industry applications
nuls-ai-service · nuls-compute-pool · API · indexer · governance tools
transparent / privacy / system tx · contract · application modules · WASM / EVM / SVM / Move
POC / AIPOW / PBFT / custom consensus · P2P · relay · evidence · checkpoint
storage · state root · cryptography · serialization · address · NA amount
Dependency direction matters: a host node may assemble concrete capabilities; a stable abstraction should not depend back on its implementations.
FRONTIER MODULES / THE WHY BEFORE THE WHAT
These modules are not a claim that every capability is production-ready. They are the protocol questions NULS Mars puts into code, testnets, specifications, and community review.
WHY IT MATTERS
A transparent ledger can reveal payment, commercial, and identity relationships forever. Selective disclosure lets a network preserve public validity without treating every participant’s context as public data.
PROTOCOL SHAPE
Groth16 / BLS12-381 · note · commitment · nullifier · Merkle root & witness · viewing key
CURRENT BOUNDARY
A network must pin the verification-key hash through its profile or governance; arbitrary user-carried verification keys are not an admission model. Full wallet lifecycle and circuit coverage remain production gates.
WHY IT MATTERS
Demand for AI compute is real. The hard question is whether security can ask miners to contribute verifiable compute capacity without making chain validity depend on subjective model outputs.
PROTOCOL SHAPE
GPU challenge → compute proof → score → leader / block order
CURRENT BOUNDARY
AIPOW is an isolated testnet and special-network experiment. GPU challenge evidence is its consensus input; model inference, HTTP probes, service quality, and energy accounting are separate service concerns — not fork choice, rewards, or slashing inputs.
WHY IT MATTERS
Keys and addresses stay in wallets, integrations, backups, and legal records for years. Planning for quantum-resistant primitives early makes an eventual migration a deliberate protocol decision rather than a rushed emergency.
PROTOCOL SHAPE
ML-DSA-65 / ML-KEM-768 / SHA-3 · binary V3 address: 72 bytes
CURRENT BOUNDARY
Longer V3 addresses are an explicit trade-off. A final standardized profile needs official known-answer tests, cross-implementation interoperability, independent audit, and a governed genesis profile before it can be described as production-approved.
WHY IT MATTERS
EVM, SVM, WASM, and Move carry different programming models, asset semantics, and developer ecosystems. A future-facing base protocol should let a network compose these worlds at explicit, inspectable boundaries.
PROTOCOL SHAPE
nuls-vm-core registry · EVM / SVM / WASM / Move capabilities · network VM profile
CURRENT BOUNDARY
A module’s presence does not claim that its runtime is activated or production-ready. Each VM needs its own safety, compatibility, governance, and admission evidence; a ChainBox profile makes activation deliberate rather than implicit.
CHAINBOX / CHAIN SPEC COMPILER
ChainBox v2 is a host-side chain-spec compiler and runtime orchestrator. It selects, generates, validates, assembles, and starts — while node crates retain the state machine, consensus, sync, and execution responsibilities.
Identity, native asset & economics, genesis consensus, VM capabilities, node topology, storage, observability, cross-chain peers, and statically linked application modules.
It does not copy the state machine, accept unaudited dynamic modules, place secrets in configuration, or fake a protocol upgrade with a local setting.
MODULE MAP
Each module below exists to make one boundary explicit. Together, they map the current NULS Mars workspace and its implementation conversations.
Common types, cryptographic suites, storage, module lifecycle, and protocol versioning.
Accounts, ledger state, transaction validity, blocks, privacy, and multisignature semantics.
P2P, node lifecycle, consensus abstractions, and distinct consensus implementations.
Contract host, VM registry, and pluggable execution capabilities for different ecosystems.
AI service, compute pools, cross-chain boundaries, and public or adapter APIs.
Chain management, ChainBox specifications, SDKs, and the command-line surface for building and operating networks.
READING THE SYSTEM
AN HONEST REFERENCE
Read each design document as a contract of intent and boundary. Implemented code, testnet support, audited capability, and production admission are separate states that should never be collapsed into one marketing statement.