Changelog and Corrections

The unbonding durations on exit liquidity have been corrected and sourced. The guide gave the middleware escrow as about seven days. EigenLayer's own specification, ELIP-002, sets WITHDRAWAL_DELAY at 100,800 blocks — about fourteen days — as a single global parameter set by governance, and states that the stake remains slashable for the whole window. The seven-day figure appears to be one component of that window rather than the window itself: EigenLayer's published rationale describes seven days as the period in which services sync to its view of stake, with the remaining week covering a fraud proof and the window to cancel an incorrect one. The stated total of roughly 8 to 14 days was therefore shorter than the first stage alone, and has been replaced with a floor and an explanation of why no ceiling can be given. The validator exit queue is now described by its mechanism instead of a day count, since its depth is live data this site does not publish. The same correction was applied to the arbitrage-delay step on LRT depegs.

Three claims have been withdrawn or narrowed after a sourcing review. One is removed. The guide on restaking in 2026 and the homepage both described a market test in which a major liquid staking token made restaking optional and most capital declined to opt in, and drew from it that institutional holders avoid restaking. Repeated searches found no published record of that outcome. Under our editorial method an unverified claim is omitted rather than hedged, so the passage and the conclusion built on it are gone. The section now makes only the point that can be shown: that staked capital is not restaked by default, and the extra reward is payment for conditions the base asset does not carry. One is narrowed and sourced. Both pages stated flatly that penalties are enforced on-chain. Eigen Labs' own announcement is now cited, and the claim carries the qualifier that announcement attaches: slashing went live in April 2025 and is opt-in, reaching an operator and its delegators only where that operator has joined a slashable operator set. The previous wording implied it applies to restaked capital generally. One loses its numbers. The guide on slashing risk said professional operators frequently manage dozens of services across thousands of delegators. Those are three magnitude claims we had not sourced. The passage now states the structural condition without them, since the correlated-failure argument does not depend on the counts.

The guide on LRT buffer and exit lessons described a liquid restaking exploit in which, it said, losses moved through individually audited contracts whose runtime interaction created an unmodelled vulnerability surface, and concluded that combining individually audited contracts does not guarantee an audited system. Checked against the published record of the April 2026 rsETH bridge exploit, that account was wrong in substance. The parties state the opposite: no contract was broken and every contract in the path behaved exactly as written. The failure was a single-verifier trust setting and the off-chain infrastructure beneath it, not an interaction between contracts. The passage has been rewritten to the sourced facts, and the lesson restated as the one the event actually supports — that a contract audit does not reach integration configuration, infrastructure single points of failure, or behaviour when an off-chain dependency is compromised. The event was also previously referred to without being named or sourced, which our editorial method does not permit; it is now named, dated and cited.

Added incident case files: dated reconstructions of documented restaking failures, written from accounts published by the parties involved, with the points they still dispute recorded rather than resolved. The section opens with three files, on the April 2026 rsETH bridge exploit, the September 2024 Penpie exploit and the April 2024 ezETH liquidation cascade. The conditions a case file has to meet are set out on the section's index page, and are strict enough that several widely-reported losses do not qualify — usually because no party published an account, leaving only secondary reporting that disagrees with itself. Case files now appear in the feed alongside the guides.

Anonymous visitor statistics were added, using Umami running on our own server. It sets no cookies, stores no IP address, and builds no profile. The privacy notice now describes exactly what is recorded, the legal basis, and how to object or switch it off. The editorial charter previously said the site used no analytics at all; that statement has been corrected rather than left standing.

The guide on slashing risk stated that most slashing events stem from software or operational glitches. That is a claim about how slashing events are distributed, and we have no public dataset attributing causes across services to support it. Under our editorial method, an unverified claim is omitted rather than hedged. The passage now states what the section actually demonstrates: that malice is not a precondition, because enforcement is mechanical and ordinary software and infrastructure failures breach the rules just as effectively. The three failure paths described below it are unchanged.

Removed the reference to the European Commission's Online Dispute Resolution platform from the Impressum. That platform was discontinued on 20 July 2025, when Regulation (EU) 2024/3228 repealed Regulation (EU) No 524/2013. Directing readers to a service that no longer exists is misleading, so the reference and its link were withdrawn. The separate statement on consumer arbitration under the German Consumer Dispute Resolution Act (VSBG) was not affected and remains in place.