StableGuard is autonomous cross-chain stablecoin depeg protection. When UST/Terra collapsed, $40B disappeared with no automated warning system — protocols kept accepting bad collateral because nothing was watching fast enough. StableGuard cross-checks two independent data sources before acting — a Chainlink price oracle and on-chain DEX data — avoiding the false alarms that plague single-feed monitoring bots. When confident a depeg is real, it automatically fires a Chainlink CCIP alert that can broadcast to multiple destination chains at once, triggering protective action before a human even notices. It doesn't stop a stablecoin from collapsing if it's genuinely dying — nothing can — but it stops the collateral damage from spreading to protocols and positions that had nothing to do with the original coin.
Price Feeds and Uniswap as two independent sources — only acting when both agree a depeg is real, avoiding the false alarms that plague single-feed monitoring bots. When confident, it automatically fires a Chainlink CCIP alert that can broadcast to multiple destination chains simultaneously, pausing new exposure before a human even notices — with fault isolation so one failed destination can't block the others.
Built for lending protocols and vaults holding stablecoin collateral without a dedicated risk desk watching prices 24/7. Pausing blocks new deposits specifically, while existing users can always withdraw — no one gets trapped mid-position.
During this event we found and fixed a real security vulnerability (performUpkeep was trusting caller-supplied data instead of live-computed values), deployed to Robinhood Chain testnet, and wired a working pause integration end-to-end against a real vault contract — not a stub.
Update: root-caused and fixed two CCIP delivery bugs (incorrect router address, missing destination gas limit) through direct on-chain forensics — decoding router events rather than guessing. Verified the full pause mechanism correct via a local fork test that proves AlertReceived, ProtectionActivated, and vault.paused() all fire end-to-end. Live testnet CCIP delivery timing remains a separate, ongoing piece.
Deployed and wired StableGuard on Arbitrum Sepolia, extending to Robinhood Chain testnet for track eligibility. The contract broadcasts CCIP alerts to multiple destinations simultaneously, with fault isolation so one failed lane can't block others — though the Arbitrum-to-Robinhood-Chain CCIP lane itself isn't live in Chainlink's directory yet, since Robinhood Chain only recently launched with Chainlink as its native oracle provider. Built and wired a real MockUSDC/MockVault integration so the pause mechanism calls pause() on an actual contract, not a stub. Found and fixed a genuine security vulnerability in performUpkeep — it was trusting caller-supplied data instead of live Chainlink feeds. Diagnosed and resolved two separate CCIP delivery bugs through direct on-chain forensics. Chainlink Automation is being sunset in favor of the newer Chainlink Runtime Environment (CRE) — we already have CRE Early Access and see migrating to it as the natural next step once Robinhood Chain's infrastructure catches up. Root-caused two CCIP delivery bugs via direct forensics: an incorrect router address, and a missing gas limit that silently starved destination execution. After fixing both, verified the full pause mechanism correct using a Hardhat fork test that impersonates the CCIP Router and calls ccipReceive directly — confirmed AlertReceived, ProtectionActivated, and vault.paused() all fire correctly (61,810 gas). Live CCIP delivery on testnet remains separately in progress.
Self-funded / bootstrapped. Actively looking for early feedback and potential integration partners rather than fundraising at this stage.
