Share
Collateral settlement is one of the most operationally demanding processes in institutional finance. A single settlement failure, a pricing discrepancy discovered at the wrong moment, a margin call that cannot be serviced before a deadline, or a cross-chain transfer that stalls in a bridge queue can trigger cascading losses that spread well beyond the original counterparty.
Traditional settlement infrastructure was designed for a world in which humans could intervene at each checkpoint. The Chainlink Runtime Environment, known as CRE, was designed for a world in which no human is in the loop at all.
This guide explains what CRE is, how its architecture maps onto collateral settlement workflows, and how to build a production-grade settlement system on top of it, with a concrete example drawn from the delivery-versus-payment use case that Chainlink’s own developer documentation uses as a reference implementation.
CRE is an offchain computation and orchestration layer that connects traditional data sources, APIs, and blockchains into a single programmable workflow. It is not a blockchain, but it is the infrastructure that sits between blockchains, financial APIs, and smart contracts, coordinating the sequence of actions required to complete a multi-step financial operation without human intervention at each step.
CRE operates through workflows, which are programs written in Go or TypeScript that define a sequence of triggers, computations, and actions. A workflow can be triggered by a schedule, an onchain event, an HTTP request, or a price threshold breach. It can read from external data sources, perform computations, call smart contract functions, and write the results of those computations onto one or more chains.
The Chainlink Runtime Environment (CRE) is powering orchestration for secure and scalable agentic workflows to unlock next-gen applications using crypto & AI.
How 7 leading platforms are leveraging CRE for AI-powered workflows ↓… pic.twitter.com/Xsyqlj39wA
— Chainlink (@chainlink) November 7, 2025
The CRE node types relevant to collateral settlement are as follows.
An EVM Log Trigger fires when a specified onchain event occurs, such as a margin call contract emitting a collateral-deficient signal. This is the entry point for automated settlement workflows. An HTTP Capability reads data from offchain sources. In a settlement context, this pulls real-time pricing from a custodian API, a bank’s payment confirmation endpoint, or a central securities depository.
An EVM Read queries the current state of a smart contract to check existing collateral balances, verify the status of a prior transfer, or confirm that a delivery obligation has been fulfilled. An EVM Write submits a signed transaction to a smart contract to release collateral, trigger a payment instruction, or update a settlement state machine. A CCIP Send initiates a cross-chain transfer through Chainlink’s Cross-Chain Interoperability Protocol, moving assets from a source chain to the settlement chain where the obligation exists.
Collateral settlement has three properties that make manual or semi-automated infrastructure inadequate at scale.
The first is time-sensitivity. Margin calls must be serviced within defined windows, often intraday or within hours. A workflow that requires human approval at each step cannot consistently meet those deadlines across a large portfolio.
The second is data dependency. Every collateral decision depends on current market prices, current account balances, current haircut schedules, and current regulatory eligibility. Stale data produces wrong decisions. CRE’s Data Streams capability delivers sub-second, pull-based price updates from Chainlink’s oracle network, stamped with a cryptographic signature that the receiving smart contract can verify onchain before acting.
The third is auditability. Regulators and counterparties require a complete, tamper-resistant record of every decision and action in a settlement sequence. CRE workflows emit structured logs at each node. Every EVM Write produces an onchain transaction that is permanently recorded. The combination gives compliance teams a reconstructible audit trail from trigger through settlement without relying on offchain databases that can be altered after the fact.
Delivery-versus-payment, known as DvP, is the settlement mechanism that ensures the transfer of securities and the payment of cash occur simultaneously and atomically. If the security does not transfer, the cash does not move. If the cash does not arrive, the securities are not released. The atomicity constraint eliminates counterparty risk at the moment of settlement.
Delivery vs. Payment (DvP) settlement across two separate blockchains is a fundamental leap forward for the global financial system.
Made possible by CCIP.… https://t.co/ez3zj2fBri
— Chainlink (@chainlink) September 15, 2023
Building DvP on CRE requires five components: a trigger contract on the seller’s chain, a CRE workflow that orchestrates the entire sequence, a bank payment verification endpoint, a settlement contract on the buyer’s chain, and CCIP for cross-chain asset movement.
The workflow begins when the seller’s tokenized securities contract emits a settlement initiation event. The CRE workflow detects this through its EVM Log Trigger and immediately begins the verification sequence.
The first action is an offchain check via the HTTP Capability, calling the buyer’s bank API to confirm that payment has been initiated and matches the agreed amount. Bank API credentials are stored in CRE’s secrets management layer and never exposed to the public chain, running inside the Decentralised Oracle Network’s trusted execution environment.
The second action is a real-time pricing check via Data Streams. Before releasing collateral, the workflow fetches the current market price of the tokenized treasury to confirm the collateral value still meets the agreed haircut threshold. This step prevents settlement from proceeding if a price move since initiation has left the collateral undersized relative to the obligation.
The haircut validation is a deterministic computation: if the current price multiplied by the collateral amount falls below the required coverage ratio, the workflow stops and triggers an alert rather than proceeding to transfer.
If both the payment confirmation and the haircut check pass, the workflow calls CCIP to move the tokenized treasury from the seller’s chain to the buyer’s chain, where the buyer’s custody account resides. CCIP handles the cross-chain transfer atomically, with the Risk Management Network monitoring the transfer in parallel and enforcing rate limits that cap potential losses from any single operation.
Once CCIP confirms delivery, the workflow writes a final settlement confirmation back to the seller’s contract, releasing the payment lock and updating the onchain settlement state. The entire sequence from trigger to confirmed settlement runs without a single human decision point.
CRE provides a local simulation environment that runs the full workflow against mock data before any transaction touches a live chain.
The simulator emits a step-by-step trace showing the output of each node, the conditional logic at the haircut check, and the parameters that would be submitted to each EVM Write and CCIP Send call. Failures caught in simulation cost nothing. Failures caught on a live chain can cost considerably more.
After simulation, deploy to a testnet such as Sepolia for end-to-end testing with live oracle data before promoting to mainnet. CRE’s deployment command packages the workflow, registers it with the DON, and returns a workflow identifier that can be monitored through the CRE monitoring dashboard.
Monitoring covers trigger latency, computation output at each node, transaction submission status, and CCIP message delivery confirmation, giving operators visibility into every step of a live settlement run.
DTCC’s Collateral AppChain, which processes approximately $4.7 quadrillion in annual securities transaction volume, uses CRE for pricing, valuation, margining, and settlement automation, with go-live scheduled for Q4 2026. Lombard Finance routes over $1 billion in Bitcoin-backed assets across chains through CRE orchestration.
Fidelity International’s FILQ tokenized money market fund uses CRE to deliver real-time net asset value data onchain via Chainlink Data Streams, with JPMorgan independently approving the daily NAV as a second verification layer. Each deployment runs the same architectural pattern the DvP example illustrates: an onchain trigger, an offchain data verification step, a conditional computation, a cross-chain action through CCIP, and a final settlement confirmation written back to a smart contract.
The key difference between a prototype and a production deployment is the secrets management layer. Bank API credentials, custodian authentication tokens, and institution identifiers must never appear in workflow configuration files.
CRE’s secrets API stores these values in encrypted form and injects them at runtime inside the DON’s trusted execution environment, ensuring that sensitive credentials are never visible to the public chain or to other workflow operators on the same network.
CRE is not a replacement for legal documentation. A DvP settlement workflow executes the mechanical transfer of assets. The legal agreement that defines what constitutes valid payment, what the haircut schedule is, and what constitutes a settlement failure remains an offchain instrument that the workflow’s parameters must reflect accurately.
If the legal agreement changes and the workflow parameters are not updated, the workflow executes correctly against its own parameters while being incorrect against the actual commercial terms. Keeping the workflow in sync with evolving legal agreements is an operational discipline that CRE does not enforce.
CRE is also not a custody solution. EVM Write calls submit transactions to smart contracts, but the private keys that sign those transactions must be managed separately through a hardware security module, a multi-party computation key management system, or a qualified custodian’s signing infrastructure. For institutional deployments, the signing infrastructure is typically the most complex part of the operational setup and sits entirely outside CRE’s scope.
For developers building collateral settlement systems today, CRE’s skill is available through the smartcontractkit/chainlink-agent-skills repository.
The skill guides compatible AI coding assistants through workflow design, capability selection, simulation, testing, and deployment, compressing the time from concept to a working DvP prototype into a single focused development session.
Share
preferred on
