Risk management in DeFi has always been a paradox. The protocols themselves are transparent -- anyone can read the smart contracts, anyone can verify the state on-chain. But the systems that monitor those protocols, the bots that trigger liquidations, the scripts that rebalance portfolios -- these have historically been black boxes. Operators run them on private servers, with proprietary logic, and users simply have to trust that they are behaving correctly.
NexusForge is changing that. Over the past six months, we have worked with three DeFi teams that are deploying verifiable agents for mission-critical risk management operations. Each team has a different use case, but they share a common requirement: the ability to prove, not just claim, that their automated systems are doing what they say they are doing.
Case Study 1: Liquidation Monitoring at Meridian Finance
Meridian Finance is a lending protocol with over $340M in total value locked across Ethereum and Arbitrum. Their liquidation system previously relied on a set of keeper bots running on AWS, monitored by an internal dashboard that only the engineering team could access. When a large liquidation event occurred in January 2026, community members questioned whether the keepers had acted optimally or whether certain positions had been selectively protected.
"We realized that transparency at the smart contract layer means nothing if the off-chain systems that interact with those contracts are opaque. Our users deserved better."
-- Elena Vasquez, Head of Engineering, Meridian Finance
Meridian migrated their liquidation bots to NexusForge agents in February. Each agent runs inside an attested TEE enclave, and every liquidation decision -- including the inputs it evaluated, the thresholds it checked, and the action it took -- produces a zero-knowledge proof that is posted on-chain. Community members can now verify that every liquidation was triggered by genuine undercollateralization, not by operator discretion. Since deployment, Meridian has processed over 12,000 verified liquidations with zero disputed actions.
Case Study 2: Portfolio Rebalancing at Vertex Capital
Vertex Capital manages a suite of on-chain structured products that automatically rebalance across yield opportunities on Ethereum, Base, and Solana. Their previous rebalancing engine ran on a centralized server, and LP token holders had no way to verify that rebalances were occurring according to the published strategy parameters.
With NexusForge, Vertex deployed a fleet of rebalancing agents that operate under a strict manifest. The manifest defines the target allocation percentages, the rebalance triggers (time-based and threshold-based), and the set of approved venues. Every rebalance action generates a proof that attests to the inputs the agent observed (token prices, pool APYs, gas costs) and demonstrates that the action conformed to the manifest constraints.
"Our LPs don't have to trust us anymore -- they can verify every rebalance against the published strategy. That's a fundamentally different value proposition for on-chain asset management."
-- Marcus Adebayo, Co-founder, Vertex Capital
Since migrating to NexusForge in late February, Vertex has reported a 28% increase in new LP deposits, which they attribute directly to the enhanced transparency. Their agents have executed over 1,400 cross-chain rebalances, each accompanied by an on-chain proof.
Case Study 3: Bridge Security at Conduit Network
Conduit Network operates a cross-chain bridge connecting Ethereum, Polygon, and Avalanche. Bridges are high-value targets -- billions of dollars have been lost to bridge exploits -- and Conduit's security team needed a monitoring layer that could detect and respond to anomalies faster than human operators while providing an auditable trail of every decision.
Conduit deployed NexusForge agents as a secondary validation layer on their bridge. These agents independently verify every bridge transaction by checking the source chain state, validating the Merkle proofs, and comparing the transaction parameters against known patterns of exploit behavior. If an agent detects an anomaly, it triggers a circuit breaker that pauses the bridge and emits a detailed proof of the suspicious transaction for human review.
In six weeks of operation, Conduit's NexusForge agents have validated over 84,000 bridge transactions and flagged two transactions that turned out to be edge-case bugs in a third-party relayer. Both were caught before any funds were at risk.
The Compliance Advantage
Beyond the operational benefits, all three teams highlighted compliance as a key driver for adopting verifiable agents. As regulatory frameworks for DeFi mature in the US, EU, and Singapore, protocols are increasingly expected to demonstrate that their automated systems operate within defined parameters. Traditional audit logs -- timestamped records stored on a database -- can be tampered with. Zero-knowledge proofs cannot. A NexusForge execution proof is a cryptographic commitment that a specific computation occurred with specific inputs and produced a specific output. No amount of post-hoc editing can forge that commitment.
For DeFi teams preparing for a regulatory environment that demands accountability, verifiable agents are not just a technical improvement -- they are a compliance strategy. And the teams that adopt them early will have the longest track record of provable behavior when regulators come asking questions.
If your team is building risk management infrastructure for DeFi and you want to explore how verifiable agents can strengthen your security and compliance posture, reach out to our solutions team or visit our documentation to get started.