zkLink Nova: Tracing the Ethereum Withdrawal

SentinelTX’s examination of the zkLink Nova withdrawal incident follows one Ethereum signer from preparation to a single transaction that paid ETH and eleven token types, then through conversions and dispersal to three recipient accounts. The execution sequence is visible in the report, but the forged-root explanation, wallet controllers and incident-attributable share of later holdings are not established.

The report also contains conflicting confidence labels and material amount warnings. This article preserves those qualifications: the observed DAI balances are not presented as traced cash-out, and the FixedFloat-labelled destination remains an unverified lead rather than a confirmed exit.

1. Scope and roles before the numbers

The seed account is the wallet at the centre of the examined withdrawal, not an identified person. Contract roles below distinguish what the call trace shows from identities or functions supplied by public reporting.

Label / Record

• Network and collection: Ethereum mainnet (chain ID 1) only. Collection: 29 October 2026, approximately 13:24–13:31 UTC; chain tip moved from block 26298676 to 26298706. Other chains were not screened.
• Seed account: 0xaca0655a11200a80d33d0e48786317304cde5718. An externally owned account (EOA): signer of the examined transaction, withdrawal recipient and disperser. Its real-world identity is unknown.
• Execution contract: 0xd5f162b68a0281686e8dc76db8d67d2f3e6aff07. Created by the seed account and called in the examined transaction.
• Payout contracts: 0x5fd9f73286b7e8683bab45019c94553b93e015cf paid ETH and was consulted before token transfers; its core/escrow role is an inference. 0xad16edcf7deb7e90096a259c81269d811544b6b6 supplied eleven token types; its zkLink Nova ERC20 bridge identity is reported, not label-verified.
• Recipient shorthand: 0x211b…9c6a = 0x211b4aebefea02470587286b398ed28fb7e09c6a; 0x75dc…c0f3 = 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3; 0x2d5a…afb0 = 0x2d5a8aabf7af4022e314eb2a54753ba2402bafb0. Similar behaviour does not establish common control.
• Meaning of an amount: A receipt or transfer is a movement; a balance is a snapshot. Gross movement counts value again at each hop. None is automatically the incident loss, a cash-out total or a recoverable amount. Unlike assets are not added together.

Public reporting collected by SentinelTX described a forged batch root injected through a public relayer and attributed detection to Blockaid. The report records estimates from USD 778,625 to about USD 786k; the USD 805k upper bound in the referral did not appear in its search results. These are reported estimates, not a loss calculated by the investigation. Only search snippets were read, not publisher pages, Blockaid’s primary publication or the Defimon Telegram channel.

2. From preparation to the withdrawal

All event times below are UTC block timestamps. The report describes a short wallet lifecycle on 28 October 2026:

• 15:49:11, block 26292232: the seed received a reported 0.02723853 ETH from a database-labelled FixedFloat address. The funding hash is not cited in the narrative, and the amount is unverified.
• 15:59:59–17:49:11: preparation included a reported 0.00045-ETH swap for 1.045988 USDC, a zero-value USDC call, a reported deposit of 0.00032933333342 ETH plus 1.0 USDC to the ERC20 bridge, and a reported 0.0021-ETH payment to a contract later seen in the call path. The preparation amounts carry the report’s unverified-amount qualification.
• 18:51:11, block 26293139: the seed created its execution contract at nonce 4.
• 22:21:35, block 26294188: the seed called that contract in position 0 of the block. The transaction succeeded, carried zero native value, used 1,109,471 gas and paid a fee of 0.001161701384313756 ETH. The report records an internal payment of 22.431294525563808989 ETH to the seed plus eleven token types. [E01]
• The seed then sent itself a zero-value transaction at position 1 of the same block. The narrative gives its fee as about 1 ETH, an unverified amount. A builder-priority-payment interpretation is the report’s assessment, not established intent.

The main conclusion treats the withdrawal as block-anchored because its hash is not cited in that section’s prose. Appendix A does list the full seed-transaction hash; it is retained in the evidence register below. That additional reference does not establish whether the withdrawals were authorised.

3. What the execution sequence actually shows

The call trace has 100 frames, a maximum depth of 9 and no revert. The seed’s contract first read the ETH-paying contract, then made two relay-path calls through several contracts. Those calls reached the ETH-paying contract before the payout. Next came the ETH payment and eleven calls to the ERC20 bridge. Each token call consulted the ETH-paying contract before ending in a token transfer. [E01]

Diagram 1. Simplified execution order within the seed transaction. Solid arrows indicate sequence, not a separately proven funds route. A forged batch root, bypassed checks and a public-relayer identity are not established.

Color and line key: slate marks observed execution stages; amber marks claims not established by the record. Solid arrows indicate execution order.

Address/role key: seed = 0xaca0655a11200a80d33d0e48786317304cde5718; execution contract = 0xd5f162b68a0281686e8dc76db8d67d2f3e6aff07; ETH-paying contract = 0x5fd9f73286b7e8683bab45019c94553b93e015cf; ERC20 payout contract = 0xad16edcf7deb7e90096a259c81269d811544b6b6. The role of 0x0fdc28e64772a70109d10fd068f859cfa7dea2c6 as a public relayer remains reported, not established.

Source: SentinelTX native PDF, CASE-ASYNCEA1, pp. 14–15; seed control record on p. 35.

This ordering has the shape of a message or root submission followed by withdrawal finalisation, which is the report’s inference. No selector, event log or calldata was ABI-decoded. The trace therefore does not establish that a root was submitted, that it was forged, or which validation checks were bypassed. No owner-signature execTransaction calls were present; the seed account, not the bridge or protocol, signed the transaction.

4. Token receipts and the conversion sequence

The receipt table records the following eleven token types. Quantities are reproduced at the table’s precision, without adding unlike assets or turning them into a new USD total:

• pufETH: 245.73789885106106. The two described conversion legs were 132.0 pufETH to the null address with 141.9532103998426 WETH received, then 113.73789885106108 pufETH to a counterparty contract with 123.264115827052162383 ETH returned. Both pufETH leg quantities are flagged unverified. [E02] [E03]
• USDC: 18,113.448053 received. The later conversion sent 18,113.494041 USDC and returned 18,109.0636110923 DAI. The report explains the larger outgoing USDC amount as the receipt plus 1.045988 USDC bought during preparation, less the 1.0 USDC deposit. [E04]
• WBTC: 0.17360197 received, followed by a conversion returning 14,048.682073834 DAI. [E05]
• Renzo Restaked ETH: 12.569253862270157, forwarded to 0x211b…9c6a at 22:55:35. Its subsequent conversion is possible but unverified.
• rsETH: 6.86647159999; STONE: 6.299750644063014; USDT: 3,360.898152; rswETH: 0.500000001; wstETH: 0.4659585510407948. The report lists sales through MetaMask Swap between 22:42:35 and 22:44:47, but does not itemise their proceeds.
• mswETH: 2.4826191990067406; mstETH: 2.6372057700999996. These were not moved in the report’s account and remained at the seed at collection.

There is a source inconsistency in the pufETH receipt section: its table gives 245.73789885106106 pufETH, while the opening sentence separately describes an unverified receipt attributed as 132.0 pufETH. Both statements remain in the PDF; this article does not substitute the smaller figure for the table total or silently reconcile them.

The USDC and WBTC conversions are described as producing 32,157.7456849263 DAI. The seed forwarded a reported 32,157.745684926304 DAI to 0x75dc…c0f3 at 23:03:59; that forwarding amount is flagged unverified. A 22:57:35 WETH-contract call is consistent with an unwrap by its selector, but its amount is not established. Two further zero-value pufETH calls were not decoded.

5. Three recipients, with amount and control limits

The chronology labels three native-ETH dispersals as signer-verified. The conclusion and warning appendix nevertheless mark their quantities unverified. The entries below therefore retain the report’s numbers as qualified source figures, not independently confirmed amounts:

• 22:58:35, block 26294373: 101 ETH to 0x211b…9c6a; amount unverified in the conclusion. [E06]
• 23:00:11, block 26294381: 100 ETH to 0x2d5a…afb0; amount unverified in the conclusion. [E07]
• 23:01:23, block 26294387: 102.061003174173709298 ETH to 0x75dc…c0f3; amount unverified in the conclusion. [E08]

The narrative’s gross total is 303.061003174173709298 ETH, subject to the same component-amount qualifications. The recipients then converted ETH into DAI through MetaMask Swap and Uniswap during the first hour of 29 October. Several swaps lack cited hashes or established inputs. Similar timing and sizes support the report’s description of splitting, but do not identify controllers or prove structuring intent; slippage management is an alternative explanation given in the PDF.

At 0x75dc…c0f3, the report describes roughly 270,847.79 DAI of inflows and 49,000 DAI leaving: a flagged 24,000-DAI MetaMask movement plus an unresolved approximately 25,000-DAI candidate leg. At 0x2d5a…afb0, the narrative includes an unverified 5,000-DAI transfer and two MetaMask movements of 4,895 DAI and 11,750 DAI. The approximately 25,000-DAI and 5,000-DAI onward legs remain unresolved in its conclusion. [E11] [E12]

6. What the recipient balances mean

The report’s balance snapshot is separate from its flow tracing. It records the following holdings at collection; no incident-attributable share was computed, and control of the three recipients is not established.

Holder / Observed balance and measurement

• 0x211b…9c6a: 268,344.091493 DAI; 0.012780831346009828 native ETH.
• 0x75dc…c0f3: 221,847.790589 DAI; 0.304521672474179225 native ETH.
• 0x2d5a…afb0: 211,962.663173 DAI; 0.293706861952877501 native ETH.
• Same-asset totals: 702,154.545255 DAI; separately, 0.611009365773066554 ETH. These are observed balances; no case-attributable share was computed and address control is not established.
• Seed account: 0.000927 ETH at approximately 13:26 UTC; 2.482619 mswETH and 2.637206 mstETH. These are the PDF’s displayed balance figures, not the higher-precision receipt quantities.
• Measurement boundary: DAI contract: 0x6b175474e89094c44da98b954eedeac495271d0f. DAI read at approximately 13:30 UTC on 29 October 2026; token-balance block not recorded. The three recipient ETH balances are at block 26298706. Source: PDF pp. 10, 21–22.

Figure 2. Observed DAI balances read at approximately 13:30 UTC on 29 October 2026. The token-balance block was not recorded. These balances are not traced proceeds, attributed cash-out or an established recoverable amount.

Color key: slate bars show DAI quantities on a common linear zero baseline. Labels preserve the exact reported balance values.

Address/role key: 0x211b…9c6a = 0x211b4aebefea02470587286b398ed28fb7e09c6a; 0x75dc…c0f3 = 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3; 0x2d5a…afb0 = 0x2d5a8aabf7af4022e314eb2a54753ba2402bafb0. Recipient control is not established.

Source: SentinelTX native PDF, pp. 10, 21–22 and 37. The PDF flags the aggregate 702,154.545255 DAI against a structured traced ceiling of approximately 270.8K DAI; it also explicitly describes the larger number as a commingled destination balance, not attributed cash-out.

The warning and the balance description coexist in the source. The larger balance must not be converted into a claim that all 702,154.545255 DAI was traced from this withdrawal. Nor does the smaller approximately 270.8K DAI ceiling establish a recoverable balance, a total incident loss or the status of the separate ETH exchange lead.

7. The FixedFloat-labelled lead is not a confirmed exit

The report describes a path from 0x75dc…c0f3 through 0x05b38c2d3c66031d72c990753f464271a7fd6def to 0x4e5b2e1dc63f6b91cb6cd759936495434c7e972f, which its attribution database labels FixedFloat. It gives the two legs as 5 ETH and 4.99999664 ETH, respectively; both amounts carry unverified-quantity qualifications. The second transaction is placed at 01:58:59 UTC on 29 October, block 26295266. [E09] [E10]

The PDF’s confidence wording is inconsistent. The executive summary calls this path “anchored end to end,” and a table prints a deposit transaction hash. The cash-out confidence paragraph nevertheless calls the destination an “Unverified lead,” says no inbound hash anchors it, and states that the path cannot be confirmed. The warning appendix repeats that downgrade. This article retains the unverified-lead status and discloses the stronger wording rather than choosing one statement as a repair.

Exchange identity is label-based, not a finding about a person. The report also says the incident-derived fraction was not computed because the native-ETH account implies other inflows. The same labelled address reportedly funded the seed earlier; that earlier payment was a withdrawal from the venue, not a deposit. No other exchange-deposit leg was identified, and Uniswap or MetaMask interactions are on-chain swaps, not custodial deposits.

8. Mimic records and the remaining evidence gaps

The report excluded four apparent ETH movements after checking their full transaction records. Each was signed by a third party, 0x80d04079ffc53eb14edbeeb37279d126f45552df, carried zero native value and called a third-party contract, while the transfer listing mimicked a case transfer. Other exclusions include token-symbol mimics “ЕТН”, “DAl.” and “ERC ”, look-alike recipient addresses, unsolicited OVER receipts and small apparent ETH receipts that were not signer-verified.

That distinction matters because a similar-looking address or symbol does not identify the genuine asset or recipient. The report set aside a DAI branch attributed to a look-alike address. Whether the candidate 5,000-DAI movement was intentional or a poisoning-induced mis-send remains unresolved. [E12]

The report’s remaining boundaries are specific:

• Mechanism and coverage: one withdrawal transaction and one signer were examined. Other bridge withdrawals on 28–29 October were not screened. No cross-chain bridge deposit was identified among observed reachable deposits; this was not a screen of every chain.
• Trace depth and attribution: onward tracing reached four of five planned steps; 13 of 41 traced transactions are itemised in the retained view described by the narrative; 11 counterparties have no attribution result. The seed’s internal-transaction history was only partly readable. The deterministic cash-out propagation did not complete and was not relied on.
• Amounts and onward movement: unresolved items include the approximately 25,000-DAI and 5,000-DAI legs, the Renzo Restaked ETH disposition, the ETH inputs to 0x211b…9c6a’s swaps, the seed’s MetaMask proceeds and the WETH unwrap amount. The narrative also describes unexplained native-ETH inflows; these are not assigned to the incident.
• Measurement and valuation: token balances lack a recorded block. The report’s broader movement inventory includes repeated hops and unidentified token symbols, so it is not a loss total. Its pricing appendix says 53 movements have a USD value, with no price source recorded for 24; six symbols could not be priced.

9. Response priorities and conclusion

The report prioritises preservation of the relevant FixedFloat order records, monitoring of the recipient and lead addresses, decoded protocol events and confirmation of contract roles. It also proposes asking the mswETH and mstETH operators whether the residual seed balances can be blocked or recovered; that capability was not verified. These are the report’s proposed actions, not actions taken for this article.

Its expert opinion is that the sequence matches a bridge proof or message-forgery exploit with rapid conversion and splitting. The supporting record remains narrower: one examined Ethereum withdrawal, later movements with amount qualifications, observed DAI holdings without established control, and a disputed exchange-destination confidence label. The report’s recovery discussion depends on negotiation, legal identification or later movement to a custodial venue, not on treating the balance snapshot as secured funds.

Figure 3. The report’s recorded execution and material limits. The forged-root mechanism and wallet controllers remain unestablished; some transfer quantities are unverified. Observed balances are not attributed cash-out, and other bridge withdrawals were not screened.

Color key: slate identifies recorded execution; amber identifies unestablished claims; sage identifies amount and scope limits. The meaning is also stated in text.

Source: SentinelTX native PDF, pp. 11, 14–15, 20–23 and 37–38. Its FixedFloat confidence text treats the destination as an unverified lead despite stronger wording elsewhere; that contradiction is preserved above.

10. Evidence appendix and source limitations

10.1 Selected transaction register

The evidence IDs in the article link to transaction pages built from hashes printed in the PDF. These links are navigation aids, not a claim of a fresh explorer check. The register states the use, UTC time, block, quantity and source qualification for each selected record. It is not an exhaustive ledger.

Evidence ID and use / PDF record and limitation

• E01 — Seed execution: 28 October 2026, 22:21:35 UTC; block 26294188. Zero native transaction value; the narrative records an internal payout of 22.431294525563808989 ETH and eleven token types. Status: successful control transaction in Appendix A; the main conclusion calls the withdrawal block-anchored. Mechanism and authorisation remain unestablished. PDF pp. 4, 14–15, 21, 35. Full transaction: 0x4f3c23e75b9134e8cae5a462550daad42fbbc19242567fcc34c0a035e297462c
• E02 — First pufETH leg: 28 October 2026, 22:24:35 UTC; block 26294203. Narrative: 132.0 pufETH to the null address and 141.9532103998426 WETH received. Status: pufETH quantity flagged unverified; the routing table separately prints a verified row. PDF pp. 7, 27, 37. Full transaction: 0xc6918af5df66ff7b632216bd0d01389a888d8c28d83c2572858f85bde4b573b3
• E03 — Second pufETH leg: 28 October 2026, 22:28:47 UTC; block 26294224. Narrative: 113.73789885106108 pufETH to the counterparty contract; 123.264115827052162383 ETH returned. Status: pufETH quantity flagged unverified. PDF pp. 7, 28, 37. Full transaction: 0x123bd9627d3c06f73498bd466100a62d4f7db76783ef9cd5ce5906a1e4fe4460
• E04 — USDC conversion: 28 October 2026, 22:30:47 UTC; block 26294234. Narrative: 18,113.494041 USDC sent and 18,109.0636110923 DAI returned. Status: conversion listed; the article uses narrative precision, without substituting the routing table’s additional digits. PDF pp. 7, 25. Full transaction: 0x2ddf70327a009c55960dd4d68fe308df772ef2816ca91215b25546f5c9d306f3
• E05 — WBTC conversion: 28 October 2026, 22:33:59 UTC; block 26294250. Narrative: 0.17360197 WBTC sent and 14,048.682073834 DAI returned. Status: conversion listed. PDF pp. 7, 25. Full transaction: 0x58bbdeebc886bfab35fa83366905fdf8c370c9ba1f9ff2dc4c39f7991c8cb239
• E06 — First native dispersal: 28 October 2026, 22:58:35 UTC; block 26294373. Narrative: 101 ETH to 0x211b4aebefea02470587286b398ed28fb7e09c6a. Status: signer-verified in the chronology, amount unverified in the conclusion and warning list. PDF pp. 8, 20, 28, 38. Full transaction: 0x2522281fbc77a612193e78e27f8cddd8147c605759216b67ce40320b9c8ff56e
• E07 — Second native dispersal: 28 October 2026, 23:00:11 UTC; block 26294381. Narrative: 100 ETH to 0x2d5a8aabf7af4022e314eb2a54753ba2402bafb0. Status: signer-verified in the chronology, amount unverified in the conclusion and warning list. PDF pp. 8, 20, 28, 38. Full transaction: 0x5696e2178bf680110d927d9bf71bdf05fcc22a282ad8a15bbdb938e5f852de44
• E08 — Third native dispersal: 28 October 2026, 23:01:23 UTC; block 26294387. Narrative: 102.061003174173709298 ETH to 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3. Status: signer-verified in the chronology, amount unverified in the conclusion and warning list. PDF pp. 8, 20, 29, 38. Full transaction: 0x04c207b1b0e181fa3aa6eb5f2ee35cdd629db5d706a555ac8e0e2a26c0b0e3bc
• E09 — Pass-through leg: 29 October 2026, 01:57:35 UTC; block 26295259. Narrative: 5 ETH from 0x75dc…c0f3 to 0x05b38c2d3c66031d72c990753f464271a7fd6def. Status: amount unverified in the conclusion; the routing table separately prints a verified row. PDF pp. 9, 20, 30, 38. Full transaction: 0x8b6d9b0a7696857ae7b956e458a1999891dcd3f73236fc7d9a10703468fbe8c1
• E10 — FixedFloat-labelled destination: 29 October 2026, 01:58:59 UTC; block 26295266. Narrative: 4.99999664 ETH to 0x4e5b2e1dc63f6b91cb6cd759936495434c7e972f. Status: destination called an unverified lead in the confidence text; amount unverified; label-based exchange identity and partial case attribution. Stronger wording and verified table rows elsewhere conflict with these qualifications. PDF pp. 4, 11–12, 19–20, 24–25, 37–38. Full transaction: 0x0268cf861201bbec96285c54cf25d66d058100a4ebd81021eeab2747dbe0127c
• E11 — Candidate onward DAI leg: 29 October 2026, 02:08:35 UTC; block 26295314. Narrative candidate: 24,999.999999999996 DAI to 0xdf410542aa35dab5e1ff3f46fcbd671d5e243977. Status: unverified lead; onward movement not established, despite a verified 25,000.0 DAI row in the routing table. PDF pp. 9, 21–22, 30, 36–37. Full transaction: 0x7c6955ba4e2a9e954966975a3d63187784a75c092a7013de447260e6e1c9d3ff
• E12 — Candidate 5,000-DAI leg: 29 October 2026, 01:32:35 UTC; block 26295135. Narrative: an unverified amount attributed as 5,000 DAI to 0x791a1cff9f9431ef98ca25c2e31ecb92a8ac2f25. Status: unresolved in the narrative; the hash appears in the routing and control records. Whether this was intended or a poisoning-induced mis-send is unresolved. PDF pp. 10, 16–17, 22, 31, 36. Full transaction: 0x8f0796105a30c02fffe21ae8410217dbc385d653578ef2fa3445073356eec3e0

10.2 Source and method register

• Primary source: SentinelTX’s 41-page native investigation report, CASE-ASYNCEA1, dated 29 October 2026. Its document-integrity block prints generation time 2026-10-29T13:53:18Z. Use: source for this article’s findings, figures and evidence references. Limitation: the PDF retains its own warnings and conflicting confidence statements. The executive summary also carries the separate case reference CASE-20261029-0001; neither identifier has been silently substituted for the other.
• Narrative and exhibits, pp. 3–23: used for chronology, call sequence, balances, exclusions, assessments and recommendations. Limitation: reported public accounts, inferred roles and qualified amounts remain distinct from established execution facts. The pufETH receipt wording and FixedFloat confidence wording are internally inconsistent, as described above.
• Path Routing Evidence, pp. 23–32: used to retain transaction references and their printed fields. Limitation: some rows are marked “verified” and show readable transfer amounts even where the narrative or warning appendix calls those quantities unverified. These conflicting labels are not resolved by this adaptation; no replacement totals or attribution shares were computed.
• Methodology, pp. 33–36: one recorded 90-day query window, 31 July–29 October 2026. The 217 retained rows with readable timestamps span 28 October 15:49:11 to 29 October 07:35:59 UTC; three retained rows have no readable timestamp. A query window is not the evidenced event span. Provider-wide account lists and the rows followed by a trace are different populations; they are not interchangeable funding or balance totals.
• Control records, pp. 34–36: used for the creation and call references, including [E01]. Limitation: a selector and successful status do not alone establish storage changes, contract control, actor intent, a loss or cash conversion.
• Warnings and integrity record, pp. 36–39: the report records one downgraded claim and 19 findings flagged for review, then lists one INV-12 entry and eighteen INV-13 entries. A signed manifest identifies the compiled source; it does not remove the substantive qualifications. Exact material warning wording is retained below.

10.3 Exact source warnings

The following is the report’s own warning record, with line wrapping normalised only. Its practical consequences are the unconfirmed destination, unverified transfer quantities and prohibition on treating the full destination balance as attributed cash-out. Where other source passages disagree, those disagreements remain disclosed rather than repaired.

PDF p. 3: INTEGRITY GATE — the integrity gate downgraded 1 claim in the text and flagged 19 findings for review. See the appendix at the end of this document for the full list.

PDF p. 4, repeated inline amount qualification: [⚠ unverified amount — auto-check: a retained transfer this line names has no readable quantity, so no retained amount bounds this figure]

PDF p. 4, holdings qualification: [⚠ exceeds the traced amount — auto-check: the structured evidence traces at most ≈270.8K DAI]

PDF p. 20, opening conclusion label: unverified (auto-downgraded) (transaction-hash anchored in this section): the cited transfer of 101 ETH is unverified, as its token scale or symbol was not retrieved and

• PDF p. 37: [INV-12] Cash-out endpoints (every exchange-deposit leg in the evidence) — a row naming a confirmed freeze target / cash-out deposit address carries no inbound TX hash anchor (a label and co-occurrence cannot establish a destination) → downgraded to an unverified lead
• PDF p. 37: [INV-13] Case narrative (time order, UTC block timestamps) — the cash-out / attributed amount '702,154.545255 DAI' exceeds the structured evidence's traced ceiling (≈270.8K DAI) by 2.6× → likely a destination's commingled total deposits counted as traced taint; auto-downgraded (re-state against the traced amount)
• PDF p. 37: [INV-13] Case narrative (time order, UTC block timestamps) — the cited transfer amount '0.00045 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] Case narrative (time order, UTC block timestamps) — the cited transfer amount '0.00032933333342 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] Case narrative (time order, UTC block timestamps) — the cited transfer amount '0.0021 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] Case narrative (time order, UTC block timestamps) — the cash-out / attributed amount '4.99999664 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] Seed receipts in the seed transaction and their disposition — the cash-out / attributed amount '132.0 pufETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] Seed receipts in the seed transaction and their disposition — the cited transfer amount '113.73789885106108 pufETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3 — the cash-out / attributed amount '32,157.745684926304 DAI' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3 — the cited transfer amount '1.0 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3 — the cited transfer amount '24,000 DAI' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 37: [INV-13] 0x75dcf3395b6cbefc6b66b28d6d93f193b58dc0f3 — the cited transfer amount '25,000 DAI' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] 0x2d5a8aabf7af4022e314eb2a54753ba2402bafb0 — the cash-out / attributed amount '100 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] 0x2d5a8aabf7af4022e314eb2a54753ba2402bafb0 — the cited transfer amount '49.8 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] 0x211b4aebefea02470587286b398ed28fb7e09c6a — the cited transfer amount '101 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] 8. Recommendations — the cited transfer amount '4.99999664 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] Findings by evidence tier — the cited transfer amount '100 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] Findings by evidence tier — the cited transfer amount '102.061003174173709298 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)
• PDF p. 38: [INV-13] Findings by evidence tier — the cited transfer amount '5 ETH' names a retained transfer with no readable quantity (its token scale or symbol was not retrieved, or its amount is unusable), so no retained amount bounds it → auto-downgraded (state it against a retained transfer's amount, or as unverified)

Source qualification, PDF p. 22: “Treat these as unverified amounts until corroborated against the sealed evidence table.” The same page states that the 702,154.545255 DAI figure is an observed commingled destination balance and must not be presented as traced or attributed cash-out. Labels are database attributions, not identity findings.