Depeg Duration Resolver Changelog

Impact Notes

• Posture set membership moved to `isFragileMintPosture` / `isUnboundedMintPosture` / `isNoPrivilegedMintPosture` / `isNoPrivilegedMintChainPosture` in shared/lib/safety-score-v9/mint-posture.ts; K1's risky-minter leg, K1's severe-surge rung and the structural-class legs all read the same source, so a future posture value is classified once instead of at each site
• 14 wrapper assets annotated `none-resolved-mint` (sUSDe, sDAI, sGHO, scrvUSD, stUSDS, sBOLD, sDOLA, srUSD, sdUSD, said, steakUSDC, steakUSDT, yvUSDC, eEARN) now publish the R1 recovery anchor at the weak rung during an event, labelled `No privileged mint authority on this token (wrapped supply can still expand)`
• The strong R1 rung remains whole-of-chain only: a wrapper's parent can still print, so `none-resolved-mint` cannot satisfy the strong-anchor leg of the tier rule. Only strong anchors enter that rule, so no verdict, sealed prediction or duration stratum moves
• Structural-class stratification is unchanged: `none-resolved-mint` still confers no robustness of its own, matching the whole-of-chain definition in shared/types/core.ts

Impact Notes

• K1's risky-minter test reads the published V9 mint posture band; the concentrated and exposed bands stay the risky set, so the rule is unchanged and only its input moved engines
• A run with no installed V9 publication derives the band from the curated authority posture instead, so a degraded or held publication cannot silently drop K1's band leg; an installed publication is authoritative, including when it publishes no band for an asset
• K5 inputs are unchanged: it already read published Safety Score outputs and continues to do so
• No factor weights, severity bands, duration landmarks, or incident-lifecycle rules change

Impact Notes

• Issuer wind-down evidence and measured exit or supply stress provide earlier terminality context, while backing concentration is gated by mechanism and observed impairment
• Duration labels follow canonical incident grouping, typical ranges use the 15th-85th percentiles, and comparable histories are deduplicated by coin
• Recovered pre-lock incidents can close and resurrect within the merge window, while regime-escalating tails begin a separate incident and prediction
• DDRR review rows expose repaired and split lineage and publish expected-versus-observed horizon calibration alongside realized outcomes

Impact Notes

• Coverage classification for incidents already active when DDRv2 was enabled is floored at the DDRv2 public-contract effective timestamp
• Historical terminal evidence before that boundary becomes terminal_before_prediction, not missed_lock_terminal
• The reviewer engine advances to ddr-reviewer-v3 so cached review snapshots are rebuilt under the corrected public audit contract

Impact Notes

• Recent registry-reviewed mint-authority incidents can make K1 supply weaponization fire even when daily supply history misses the mint path
• Very-high-risk reserve concentration is elevated by default and severe only with a severe/catastrophic below-peg break or a frozen/dead dependency
• Stage 2 duration remains on duration-landmark-v1 because the DDRR duration sample is still too small and clustered for a fitted retune

Impact Notes

• Sealed live tails can be adopted when they reopen within 6 hours of the prior current source event closing
• Duplicate sealed incidents can be superseded through append-only errata and lineage while downstream reads resolve to the canonical incident key
• DDRR evaluates repaired canonical predictions against the effective current source event instead of scoring an older closed fragment as recovered
• The APXUSD June 2026 duplicate prediction is invalidated and treated as an alias of the unresolved June 2 incident

Impact Notes

• Uses fresh DEWS sub-signals to derive bank-run and blacklist-surge inputs for K5 and K3
• Uses the same 7-day DEX TVL baseline selection as the liquidity API for K5 exit-collapse checks
• Hydrates the latest Safety Score history row so R5 mean-reversion anchors and related context use live report-card data
• Marks DDR runs degraded when these required context-source queries fail instead of scoring with silently absent inputs

Impact Notes

• Introduces the readiness-72h-v1 forecast-readiness version and a strict early-lock threshold
• Publishes row-level readiness components and reasons as forecast readiness, not a probability or confidence label
• Adds optional/defaulted lock trigger, readiness, and 72h backstop metadata to the public contract while preserving legacy rows
• Includes new immutable readiness metadata in public row hash payloads when present

Impact Notes

• Freezes one official public_prediction outcome per canonical incident key instead of recomputing public forecasts every run
• Adds pending, lock-deferred, publication-retry, no-call, frozen, and invalidated public states with live facts separated from frozen prediction payloads
• Anchors duration estimates to the lock timestamp and preserves frozen fields on stale responses while marking live overlays stale
• Moves DDRR accountability to first-published sealed outcomes and explicit coverage states rather than mutable latest snapshots

Impact Notes

• Stores quarter-hourly DDR assessment checkpoints and reviews the first checkpoint for each event under the current methodology
• Publishes recovery-likelihood accuracy and average observed-minus-DDR recovery-duration error on /depeg/ and GET /api/depeg-resolver-review
• Keeps pending, insufficient-signal, and data-issue rows visible while excluding them from scored headline accuracy
• Removes terminal-lifecycle assets from live DDR readouts while DDRR scores their stored predictions as terminal outcomes

Impact Notes

• Stage 1 emits an ordinal verdict (recovery_likely / at_risk / recovery_unlikely / insufficient_signal) from kill signals and recovery anchors over structural metadata and the live depeg fingerprint
• Stage 2 emits a depth/direction/structural-class stratified landmark estimate with per-horizon (6h/24h/7d/30d) resolution likelihood, support-gated and Wilson-bounded
• Verdicts are calibrated domain reads, not fitted probabilities; audit-verdict gating is not used because event provenance is unpopulated