Architecture

sanect is a privacy-focused EVM Layer-1 (codenamed pvtchain during early development; renamed to sanect for the public launch).
Status: mainnet live (sanect_7628-1, EVM chain ID 7628) since 2026-06-22. Shielded path (ShieldedPool v1.4) is deployed and verified end-to-end — see Shielded transfers for usage and Privacy model for the design.

Goals

Requirement	Target
Block time	~400 ms
Throughput	2,000–5,000 tx / block
Consensus	DPoS, top 50 validators by stake; open delegation
Smart contracts	Native EVM (Solidity, MetaMask / JSON-RPC)
Economics	Custom token denom, custom gas model, configurable gas price
Transfers	Two paths: (1) public send/receive, (2) shielded send with passive auto-receive (no claim tx)
Nodes	Anyone can run a full node / public RPC node
Testnet host	Railway, 24 GB / 24 vCPU per box, up to 1000 boxes × 100 replicas

Stack — Cosmos SDK + CometBFT + evmOS

            EVM contracts (Solidity, MetaMask, ethers)
                            │
                ┌───────────▼───────────┐
                │     x/evm + x/erc20   │  evmOS modules
                │     x/feemarket       │
                ├───────────────────────┤
                │   x/staking (DPoS)    │
                │   x/bank x/mint ...   │  Cosmos SDK v0.53.4
                ├───────────────────────┤
                │      CometBFT v0.38   │  BFT consensus, ~400 ms blocks
                └───────────────────────┘

Why this stack:

CometBFT consensus reaches sub-second finality and is tunable to ~400 ms with a small, well-connected validator set.
Cosmos SDK x/staking is DPoS out of the box: set MaxValidators = 50, top-by-stake become active, delegation / redelegation / unbonding / slashing all built in. Minimal custom code needed.
cosmos/evm modules (x/evm, x/feemarket, x/erc20) provide a native EVM inside the Cosmos binary — real Solidity, JSON-RPC, MetaMask — while keeping custom modules and staking in the same state machine. This is the only mature path to "EVM + custom L1 modules + DPoS" in one chain.
Custom precompiles are supported — required for fast in-EVM verification of the privacy proofs.

Rejected options:

Substrate/Polkadot — weaker EVM via Frontier, significantly steeper learning curve
From-scratch chain — reinventing consensus/staking is not worth the cost for a small team

Consensus & performance tuning

CometBFT config

Validators run config.toml with:

toml

timeout_commit              = "400ms"   # primary block-time lever
timeout_propose             = "300ms"
skip_timeout_commit         = false
create_empty_blocks         = true      # keep 400ms cadence even when idle
create_empty_blocks_interval = "0s"

Genesis consensus params

json

{
  "block": {
    "max_bytes": 22020096,
    "max_gas":   150000000
  }
}

max_bytes = ~21 MB gives room for 2–5k transactions. max_gas = 150M gives headroom for ZK proof verification alongside normal transactions.

Hard truths to design around

Constraint	Implication
50 validators tolerates 16 faulty (BFT `f = (n−1)/3`)	Deliberate decentralisation / latency tradeoff — fine for a fast chain
400 ms `timeout_commit` means wall-clock RTT dominates	Spread validators across regions per Variant A topology (Asia 52% / EU+US 48%) — cross-continent latency is manageable with tuned timeouts
2–5k tx/block at 400 ms ⇒ 5k–12k TPS target	Bottlenecks will be EVM execution and mempool gossip, not consensus — plan load tests early

Performance targets

	Public (EVM / bank)	Shielded (when shipped)
Tx per block	2,000–5,000	~100–500 (ZK proof verify ≈ 1–3 ms each)
Tx size	~150–300 B	~1–2 KB (ZK proof + encrypted notes)
Client-side proving	none	~0.5–3 s per shielded spend
State growth	prunable	monotonic (nullifier set can't be pruned)

Node topology

Roles

                        ┌──────────────┐
                        │  50 Validators │  (consensus only, hardened, co-located)
                        └──────┬───────┘
                               │ private p2p
                   ┌───────────▼───────────┐
                   │    Sentry full nodes   │  (DDoS shield, 2–3 per validator)
                   └───────────┬───────────┘
                               │ public p2p
                   ┌───────────▼───────────┐
                   │  Public RPC replicas   │  (read-only, scale via state sync)
                   └───────────────────────┘

How it scales:

50 validators — the only nodes in consensus. Co-located, hardened, persistent volumes.
Sentry nodes — validators sit behind sentry full nodes (DDoS shield). Public traffic never reaches validators directly.
RPC replica fleet — public-facing read nodes that sync via CometBFT state sync. Scale horizontally without overloading validators. Front with a load balancer.
State sync enables new nodes to join without replaying the full chain from genesis.

This maps to Railway's topology: private networking for validator↔validator traffic; public subdomain for the RPC fleet.

Economics — token & gas

Token: SNCT (display) / asnct (base denom, 18 decimals)
Gas: x/feemarket (EIP-1559 style) — base fee + tip, configurable min gas price, burn/tip split set in genesis. Lets you fix a floor price and let it float under load.
Fee token for shielded txs: same denom as public txs (simplest; separating to a different denom moves to v2 if needed).
Staking rewards / inflation: x/mint + x/distribution. Parameters set in genesis.

Module layout

app/
  app.go                       # wire SDK + evmOS + custom modules
x/staking      (SDK)           # DPoS, MaxValidators = 50
x/evm          (evmOS)         # native EVM
x/feemarket    (evmOS)         # EIP-1559 gas
x/erc20        (evmOS)         # ERC-20 <-> bank denom
x/mint, distribution, slashing # rewards / security
precompiles/                   # staking, distribution, gov, slashing, bank, WERC20

The privacy module (ShieldedPool v1.4) is a Solidity contract on the EVM, not a Cosmos SDK module. It uses a Noir circuit compiled to UltraHonk proofs verified by an on-chain HonkVerifier contract.

Deployment

Component	Setup
Validators (50)	NVMe-backed hosts (Vultr, Hetzner, or bare metal), low-latency regions per Variant A topology, persistent volumes for chain data
Sentries	2–3 per validator, public-facing, shield validators from direct connection
RPC replicas	Read-only nodes scaled via state sync; fronted by load balancer
Archive RPC	Full-history node (`PRUNING=nothing`) at `archive.sanect.com` for indexers and explorers
Genesis & keys	Scripted genesis via `scripts/localnet.sh`, validator registration via staking precompile
Observability	Prometheus (`instrumentation.prometheus = true`) + Grafana for block time, mempool, missed blocks

NVMe recommended

400 ms blocks require local NVMe storage. Shared/networked storage (ZFS, EBS) introduces I/O contention that can push block times above 700 ms. See Run a validator for hardware recommendations.

What's shipped

Mainnet live (sanect_7628-1, EVM chain ID 7628) since 2026-06-22
ShieldedPool v1.4 with Noir/UltraHonk proofs — full shield/send/unshield verified
AMM DEX (Uniswap V2 fork) with 15+ LP pairs and MasterChef farm
On-chain Order Book DEX (CLOB) with limit/market orders
.snct name service (registration, auctions, marketplace, reverse records)
Hyperlane bridge to Ethereum mainnet (14 ERC-20 assets)
Browser-based ZK proving (~2-3s per transaction)

Roadmap

Multi-chain bridge expansion (Base, Arbitrum, Optimism, Polygon, BSC)
Multi-validator ISM upgrade (replace TrustedRelayerIsm with multisig)
IBC integration for Cosmos ecosystem connectivity
Off-chain note indexer for faster wallet sync
External security audit

Architecture ​

Goals ​

Stack — Cosmos SDK + CometBFT + evmOS ​

Consensus & performance tuning ​

CometBFT config ​

Genesis consensus params ​

Hard truths to design around ​

Performance targets ​

Node topology ​

Roles ​

Economics — token & gas ​

Module layout ​

Deployment ​

What's shipped ​

Roadmap ​

Further reading ​