⌘K

Live evidencecounted 2m ago

0.50days of MEME volume to unwind the crowd99%of the XMR book on one side, across 30 vaults$1.30Bunder one key with no guardian, 52 of the 424 vaults whose control we have read$334.2Mwhose control we have never read, across 3,230 vaults$1.95Bsitting in vaults and not lent out$11.27Bunder 3,654 vaults we read$4.79Bin 323 vaults where we saw automated execution, nothing declared2,080vaults declared or on a venue built to run software, holding $1.01M324software-run vaults scored, 13% of them450operators identified behind them26networks where we read capital, from Ethereum to Cardano31,361,886deposits and withdrawals read from chain5changes recorded in the last day594,038ERC-8004 agent registrations read−0.57/ 30 days364,225distinct owners behind those registrations200,313registrations carrying a readable card, 34%3,337,008paid requests on the Olas rail (deliveries are counted apart)−0.66/ 30 days87suppliers delivering that work411,027agent-to-agent jobs read on Base+0.04/ 30 days$1.0Msettled between agents, all time1,175suppliers listed on the ACP rail189agents live on Cardano76Cardano agents deregistered, 29% of all ever minted29,141escrows between Cardano agents since Jun 2025, 142 still open78%of every agent payment we read comes from 2,571 addresses that never stop307,376addresses paid once and never came back, of 494,774 that ever paid934of those payers are an agent we can name, 0.189% of them635agent hosts answered when we asked, of 2,532 probed17,125agents sit behind a host that answers nothing at all404msmedian answer from the 635 hosts that answer44%of all agent ratings were written by ten addresses, of 671,178 ratings read29,502rated agents have exactly one rater, of 47,110 rated at all86,979service endpoints declared across 33,840 agent cards we read0.50days of MEME volume to unwind the crowd99%of the XMR book on one side, across 30 vaults$1.30Bunder one key with no guardian, 52 of the 424 vaults whose control we have read$334.2Mwhose control we have never read, across 3,230 vaults$1.95Bsitting in vaults and not lent out$11.27Bunder 3,654 vaults we read$4.79Bin 323 vaults where we saw automated execution, nothing declared2,080vaults declared or on a venue built to run software, holding $1.01M324software-run vaults scored, 13% of them450operators identified behind them26networks where we read capital, from Ethereum to Cardano31,361,886deposits and withdrawals read from chain5changes recorded in the last day594,038ERC-8004 agent registrations read−0.57/ 30 days364,225distinct owners behind those registrations200,313registrations carrying a readable card, 34%3,337,008paid requests on the Olas rail (deliveries are counted apart)−0.66/ 30 days87suppliers delivering that work411,027agent-to-agent jobs read on Base+0.04/ 30 days$1.0Msettled between agents, all time1,175suppliers listed on the ACP rail189agents live on Cardano76Cardano agents deregistered, 29% of all ever minted29,141escrows between Cardano agents since Jun 2025, 142 still open78%of every agent payment we read comes from 2,571 addresses that never stop307,376addresses paid once and never came back, of 494,774 that ever paid934of those payers are an agent we can name, 0.189% of them635agent hosts answered when we asked, of 2,532 probed17,125agents sit behind a host that answers nothing at all404msmedian answer from the 635 hosts that answer44%of all agent ratings were written by ten addresses, of 671,178 ratings read29,502rated agents have exactly one rater, of 47,110 rated at all86,979service endpoints declared across 33,840 agent cards we read

Survivorship

The vaults that died, counted, because a survival rate cannot be computed from the survivors.Note 19

One venue. Every count here is Hyperliquid’s, because it is the venue that publishes its dead: the vaults it has ever listed, whether they are open, closed or left holding dust. Other venues appear elsewhere on this site and not in this population, a survival rate computed across venues that do not publish their closures would be computed from the survivors, which is the error this page exists to name.

59 days of census, one reading a day, from 12 August 2026. That is short, and it is short because we started recording it then rather than because nothing happened before. This is a reading we took and stored rather than an event a chain kept, so the series itself begins when we began; the underlying activity may well be reconstructable from an archive node or the venue’s own history, and we have not done it.

What became of each month’s intake

Three outcomes, not two. A vault is either large enough for us to rate, alive but below that floor, or closed and abandoned, and the middle one matters, because a month of mostly-small-but-living vaults is not a month where most of them died.

Reach rateable size

2.3%

216 of 9,448 born

Closed or abandoned

85.2%

8,047 vaults

Alive, under the floor

12.5%

1,185 vaults, neither outcome

Drawn from 39 cohorts at least six months old; 6 younger months are excluded. Those three add to 100%, which is why there is no separate success chart, the failure rate is not the complement of the survival rate, and the gap between them is the middle column.

rated today alive, under the rating floor closed or abandoned

Month	Born	Rated	Small	Gone	Rated %
2025-05	202	5	51	146	2.5%
2025-06	251	11	51	189	4.4%
2025-07	224

A different population from the rest of this page. The census above is Hyperliquid’s alone; this line is every platform we read, and it is about our own coverage rather than about a market. It is here because the same argument applies to us: a series that begins when we began cannot describe what happened before, and it steps when our reach steps rather than when capital moves.

Capital under measurement over time

$11.27B measured, 43% machine-managed · 12.3% of all DeFi

Machine-managedMeasured, not established19 Sept: Morpho Vault V2 added to what we read, about $4.75B that existed before1263 of 1318 days rebuilt from stored NAV · coverage stepped 12.2× at the join

The two halves are not comparable and the boundary is drawn for that reason. To its right, every day was observed. To its left, days are rebuilt from NAV readings we happened to have stored, carrying each vault’s last known value forward, so they cover far fewer vaults, and they use today’s vault list and classification, which makes them survivorship-biased backwards. The step at the join is 12.2× (337 to 623 vaults where we have the counts): that is our reach arriving, not the market growing. Read the left half as a floor under what existed, never as a measurement of it.

Method and limits in the methodology. This page is free and will stay free: it is what makes everything else here worth reading.