Global Gold
A reserve-backed gold routing asset at the center of an omnichain market system — carried from a founder’s conviction to one working testnet across four chains, with the third-party audit handoff prepared. The first six steps of the Coincraft Method are recorded here as completed work; the audit handoff into step seven is prepared.
- Testnet
- Four chains
- One omnichain mesh
- Audit handoff prepared
- Not mainnet
- Not independently audited
What the founder wanted to make possible.
Cole Chapman arrived with conviction, not a spec: gold should be able to move through onchain markets as money again. Making that real meant gold functioning as a monetary routing layer — liquidity concentrated through a gold core, a reserve-backed asset carrying it, cross-chain operation, and the governance and incentive machinery around it.
None of that could be assembled as parts. The economy, contracts, cross-chain state, liquidity design, governance logic, and demo surface had to resolve into one reviewable protocol system.
I put Coincraft’s client practice aside for a year to partner with Cole directly — serving as Global Gold’s Chief Web3 Officer and building the protocol in that role.
Once that premise held, the next question was structural: what system would make it true?
The system found its shape.
Global Gold was not designed as a single vault, a single token, or a single DEX. It was designed as a gold-native trade architecture — a monetary layer and an omnichain layer resolving into one system.
The gold core is the primary route.
Every listed asset has a gGOLD pool. For N listed assets, that hub layer needs N pools rather than the N(N−1)/2 pairs of a fully connected market. Jurisdictional gold assets are the designed exception: they also pair directly with one another, adding arbitrage routes intended to deepen volume, tighten price alignment between the golds, and create more fee flow.
The reserve underneath is a communal basket vault: multiple tokenized-gold reserves in one redemption claim, ounce-for-ounce, with oracle-free redemption accounting.
Plate 01 · The Trading Wheel. gGOLD at the center; the jurisdictional gold shell and the L1 membrane around it. Shell 1 shown at the design target: twelve slots, USG first, with PAXG and XAUT as third-party Ethereum anchors for architecture context.
Cross-chain without losing the anchor.
gGOLD is omnichain over LayerZero V2 with a lockbox home on Ethereum: bridged supply stays locked at home, so global supply is always anchored where the gold is. Governance state crosses chains as verified messages, not wrapped tokens.
Delegation reports travel in numbered caravans with per-user idempotency cursors: out-of-order delivery is absorbed, replays are skipped, recovery is timelocked. Monotonic, idempotent omnichain bookkeeping — reviewable and invariant-tested.
The pattern is designed to run on any EVM chain with Uniswap v4; the testnet proof runs on four.
Plate 03 · The Omnichain Elevator Tower. One vertical OFT shaft threads every gGOLD core, anchored to the Ethereum lockbox.
The remaining plates in the drawing series detail the architecture’s edges — select any plate to watch it in motion. Vision layers are identified on their plate faces.
Swipe to inspect the plates →
Steering was designed into the machine.
GOLDN steers liquidity incentives through commitment and delegation: longer locks carry more voting power over emissions and protocol-fee flows, while deeper delegation earns lower swap fees through a Uniswap v4 dynamic-fee hook.
With the protocol standing first, the token had real work to do — steering depth, fee bands, and emissions inside an architecture that did not depend on it to exist.
Plate 02 · The Steering Wheel. veGOLDN at the hub of the gauge ring — vote weight steering emissions, fee flows, and fee tiers.
Gold backs the routing asset. The routing asset unifies liquidity. Governance steers depth. Omnichain rails carry the structure.
The architecture became the source of truth.
The architecture entered specification as a set of decisions that could still be challenged — systems language resolved into smart-contract language, every contract, lifecycle state, permission, and invariant settled before production code began. That specification became the source of truth the build would test.
Built to catch insolvency, gate privileged access, and slow dangerous changes.
For Global Gold, step 04 specified Guard Contracts to control access to sensitive functions, slow dangerous changes, and lock the system down when something is wrong. Around them, I designed the timelocks, constrained executors, function-level allow-lists, circuit breakers, insolvency tripwires, emergency paths, and lockdown gates selected for this system’s assets, permissions, governance, and failure modes.
Where a separate guardian framework is part of the stack, I integrate it into the protocol’s control architecture. In Global Gold, I introduced the original multi-wallet guardian concept; the CTO designed and built the guardian framework itself. My work is the protocol-specific integration architecture and the surrounding control infrastructure.
Artifact A2 · The Control Architecture, drawn from the flagship security pass. Plates 01–06 document the Global Gold architecture; Coincraft artifacts A1–A2 and D1–D3 document the method and its delivery record.
Smart Contract Development & Testnet Deployment
COMPLETED · IN REVISIONThe specifications became contracts, tests, and security infrastructure — the implementation passed eight internal review rounds. One testnet was deployed across four chains and wired through twelve LayerZero V2 peer routes — rehearsed on forks, then verified end to end.Produced · contracts · tests · review record · deployed testnet · manifests and runbooksThe specification survived contact with the build.
A frozen specification is not permission to stop thinking. Implementation put the architecture under a different kind of pressure — the control surfaces designed in step 4 were built into the contracts, then armed and wired to the deployed system, and two material changes returned through the same specification, review, and verification discipline.
Built the CDP model. Retired it.
Global Gold originally included a CDP-style debt model. It reached the v1 testnet with contracts built and the frontend integrated. I retired it because it was the wrong protocol architecture. Minting against LP collateral kept the asset locked to liquidity providers and left its price without a redemption anchor. The basket vault replaced it with direct mint and redeem — open to anyone who wants to hold the asset, redeemable ounce-for-ounce against the reserve — and the economics anchored to the gold itself.
Part of finding what should exist is retiring what should not.
Bytecode headroom kept collapsing. I repackaged the architecture.
Rather than continue extracting modules to work around the EVM size limit, the protocol’s controller, cross-chain hub, and reward distributor were repackaged into hot cores and selector-routed facets sharing ERC-7201 storage. The change moved through staged CI gates, cross-review, and internal close-out. The invariant suites showed no assertion-line drift, and the vault redeem path was outside the change set.
Architecture is not frozen in place. Changes return through the same specification, review, and verification discipline as the original build.
Those decisions sit inside a larger chain of record. The implementation passed eight internal review rounds, followed by extension and security review. Findings-ledger discipline held throughout: every finding was recorded, reviewed, remediated, or explicitly carried forward.
One testnet environment. Four chains. One mesh.
The record then had to survive outside the repositories. Not four separate testnets — one environment spanning Base Sepolia, Ethereum Sepolia, Optimism Sepolia, and Arbitrum Sepolia, integrated through twelve LayerZero V2 OFT peer routes. Hub authority, delegation reporting, and tier passports route through Base; supply stays anchored to the Ethereum lockbox.
Cross-chain protocols require more than contracts deployed on different networks: routes, peers, registries, deployment manifests, chain-specific runbooks, operating ceremonies, and verification.
The protocol has to be inspectable across the mesh, not just plausible on a diagram.
Scroll horizontally for full output →
gGOLD vault solvency invariant suite — archived Foundry run, July 7, 2026. 256 runs · 16,384 calls each · 0 reverts. Fuzzing, not formal verification.
Findings were tracked.
The trail matters: finding, decision, remediation, verification, disposition.
| ID | Severity | Status |
|---|---|---|
| RE-H1 | HIGH | CLOSED |
| R14-H1 | HIGH | CLOSED |
| RE-M3 | MED | CLOSED |
| R14-M2 | MED | CLOSED |
Redacted excerpt — GOLDN findings ledger. ~135 findings tracked to closure across the spec and build audit rounds. Titles and implementation details removed.
The record gives a security team a coherent basis for review. Internal adversarial review is multi-model by design; every finding’s disposition is human-owned. How the build waves work →
Front-End Development & Optional Gamification Layer
COMPLETED · IN REVISIONThe investor dashboard and the Observatory became two connected inspection surfaces for reserves, solvency, bridge state, trading, governance, and portfolio flows. Claim portals were built.Produced · the finished frontend · claim portals · the ObservatoryThe repositories became a product people could inspect.
The dashboard at the top of this record is the system’s front door. The Observatory is its inspection room: it renders the four-chain topology and the deployed system’s state, so reviewers can inspect the system without reverse-engineering a repository.
The captures reflect the testnet wired June 10; any trading shown runs in the demo/fork environment.
Demo access is provisioned with the technical walkthrough invitation.
The product is built and inspectable. The first six steps are complete; the audit handoff into step seven is prepared.
Pre-Mainnet Prep & Mainnet Deployment
PREPARED · NOT COMPLETEDThe audit handoff package is assembled. Third-party audit, correction, and mainnet remain ahead.Produced to date · the audit handoff packageWhat this record does — and does not — claim.
- Testnet deployed and wired
- Demo-ready
- Vault solvency invariant-tested (Foundry, July 7)
- Adversarially reviewed
- Prepared for internal security review
- Prepared for third-party audit handoff
- Third-party audited
- Mainnet deployed
- Security certified
- Formally verified
- Production mainnet infrastructure
I will make each further claim only when the corresponding evidence exists.
Building a protocol of your own?
Global Gold shows what the method leaves behind: a protocol carried through six completed steps, with its handoff into third-party audit prepared. If your protocol still has unresolved decisions between the vision, the economy, the contracts, and the audit path, tell me where the system stands — the conversation starts there.
Serious prospects can also request a guided technical walkthrough of the build — an inspection of approved artifacts; repository access is not included.