Service status for the desk: data freshness, source availability and 90-day reliability. An independent monitor runs 44 checks against the production site every 15 minutes, covering availability, data integrity, onchain verification of reported liquidity, risk inputs and scoring.
Availability is the share of checked 15-minute slots outside a warning listed under Recent incidents. Slots in which no check ran are shown as not checked and excluded from the figure. Daily bars: no listed incident (green), a caution longer than an hour (amber), some slots not checked (grey), a warning of 30 minutes or more (red). Short or internal cautions are not shown. Each bar's tooltip gives the count. Incident record starts 30 Sep 2026. Earlier days show their colour only. Record starts 29 Sep 2026.
Warnings lasting 30 minutes or more, and cautions lasting longer than an hour. Times in UTC.
Aave Pro: data partly unavailable. Aave Pro: missing from the desk. Part of the data behind the page did not load, loaded only on a second attempt, was out of date, or held different figures from the full data for that time. Aave Pro did not respond, or no venue passed every check, so the source's cards and venues were not shown for that period.
Pendle: data partly unavailable. Part of the data behind the page did not load, loaded only on a second attempt, was out of date, or held different figures from the full data for that time.
Collateral risk data was missing or out of date. Some figures briefly missing or out of date. Collateral risk data behind some venue scores was missing or older than usual. Part of a data service did not respond, so some figures were missing or older than usual.
Rates were late to refresh. Some figures briefly missing or out of date. Figures were older than usual, or their timestamp could not be read. Part of a data service did not respond, so some figures were missing or older than usual.
Rates were late to refresh. The site returned errors. Figures were older than usual, or their timestamp could not be read. The site's server logged crashes, time-outs or repeated errors.
Earlier incidents and the monitoring methodology are documented in the handbook.