> For the complete documentation index, see [llms.txt](https://docs.usegimbal.app/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.usegimbal.app/risk/precedents.md). # Four precedents the design answers Four incidents, at Maple, at Goldfinch, at Centrifuge and at Edel Finance, the figures behind each one, and the rule in Gimbal that answers it. Putting a loan on chain changes how it settles. It changes nothing about whether the borrower can pay. Several protocols blurred that line between 2022 and 2026, and the record of what it cost them is precise. Gimbal treats that record as a specification. Every one of the four cases below maps to a rule in the protocol. ## Three rules, drawn from four losses Each incident traces back to one of three choices: lending with no security, depending on enforcement that happens off chain, or valuing an asset other than the one actually held. Gimbal rules all three out at the design stage. 1. **Only overcollateralised loans.** The collateral is liquid, priced by an oracle and held on chain. 2. **Only smart-contract recovery.** An auction runs without anyone's consent or cooperation. 3. **Only the escrowed token is priced.** Wrappers are never accepted, and Chainlink feeds sit behind the protocol's own checks. What remains after those three rules are issuer risk, closed-market gaps, the sequencer and the code itself. The rest of this section covers each one alongside its control. ## The cases ### Orthogonal Trading defaults on Maple Finance, December 2022 | | | | -------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Loss | About $36M, roughly 30% of everything Maple had lent | | What happened | Orthogonal Trading took uncollateralised loans from Maple's pools, misstated its FTX exposure, and defaulted once FTX collapsed. | | Lesson | Do not lend unsecured to a counterparty you cannot see into. Maple now runs at about 165% collateralisation with the collateral held in custody. | | The answer in Gimbal | No loan is written without collateral in escrow above the debt. The design never asks a borrower to disclose anything, because nothing depends on the disclosure. | ### Goldfinch borrowers default, 2023 to 2026 | | | | -------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Loss | About $18M spread over three borrowers: Tugende, Stratos and Lend East. The GFI token lost 99.8% of its value, and in June 2026 the platform entered maintenance mode. | | What happened | Borrowers in emerging markets were bound only by off-chain covenants. They defaulted, and recovery was slow and incomplete; Lend East returned about 42%. | | Lesson | A covenant cannot be enforced by a chain. Recovery off chain is a legal process that takes years. | | The answer in Gimbal | Recovery takes the form of a Dutch auction. Any address can open it, and it runs for about 45 minutes. Nothing needs enforcing because the collateral already sits in the contract. | ### ConsolFreight and Harbor Trade on Centrifuge, 2023 | | | | -------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Loss | About $4M, MakerDAO's exposure included | | What happened | Trade-finance pools were built on a small number of obligors. When those obligors failed, redemptions froze for months. | | Lesson | Concentration kills, and inside a pool a freeze hits everyone at once. | | The answer in Gimbal | Per-token caps hold concentration down. Each slice stands alone, so a bad position freezes only itself. There is no pool and no redemption queue that could be frozen. | ### The wGOOGLx exploit on Edel Finance, July 2026 | | | | -------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Loss | About $403K in bad debt | | What happened | Chainlink's price for tokenised Google stock was correct throughout. The exchange rate of a wrapper token was inflated 78 times over and accepted as collateral value. | | Lesson | Price the exact token in hand. Every derived rate is a place to attack. | | The answer in Gimbal | `PriceSentinel` points at the feed of the escrowed token and at no other source. A wrapper, a vault share or a derived rate never counts as collateral. See [Pricing, sessions and the price checks](/risk/pricing.md). | --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.usegimbal.app/risk/precedents.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.