Post-Mortem: On-Chain Analysis of the COI Addendum Vote
The [ARFC ADDENDUM] Mandatory Disclosures and Conflict-of-Interest Voting Norms Snapshot vote has closed. The proposal did not pass.
This is the final update in this series. Prior analysis: Original findings 1–4 | Findings 5 update
This post documents one additional finding and presents the complete record.
New Finding: 40,000 VP from Aave Genesis Team Infrastructure
0xD9a910aC062896d2C29cF7c11486895e84f58392 cast 40,000 VP NAY at 04:17 UTC on February 13 - hours before close. This was the wallet’s first and only Snapshot vote. Nansen labels it “Aave: Smart Account”.
This is a 3/7 Gnosis Safe created on December 13, 2024. Five of its seven signers have zero ETH and zero transactions - consistent with addresses generated specifically for this Safe.
Funding path:
The 40,000 AAVE was not sent via standard ERC-20 transfer. It was routed through Aave V3: the parent Safe supplied AAVE to the V3 Pool, then called withdraw(AAVE, 40000, targetSafe). This produces a Transfer event originating from the aAAVE contract rather than the sending Safe, which prevents standard transfer-tracking tools from linking sender to receiver.
Both V3 withdrawal transactions - a 0.1 AAVE test (tx) followed by 39,999.9 AAVE (tx) - were executed on December 17, 2024, by EOA 0x106fc088aBA908130fBC343F2F6d212Ff36150D1. This EOA is a signer on 52 Gnosis Safes (per Safe Transaction Service enumeration), including:
The Grandparent Safe shares 6 of 7 signers with the Parent Safe. Same key holders across the chain.
The wallet held 40,000 raw AAVE - never staked, never supplied, never used - for 14 months before casting this single vote.
Final Vote Result
NAY 688,619 vs YAE 603,607. Margin: ~85,000 VP. 171 total voters.
Complete Attribution Table
| Finding |
VP |
Attribution |
| 1. AAVE Multisig delegation |
333,000 |
Identical 7/7 signers with Nansen-labeled AAVE Multisig |
| 2. Stani-labeled Safe delegation |
40,880 |
Nansen: “Stani Kulechov, Public Figure” |
| 3. Whale via Aave Liquidation Account |
111,885 |
Funded via Aave Liquidation Account + kulechov.eth ENS ownership |
| 4. Fresh Safe from AAVE Multisig |
106,282 |
stkAAVE from Nansen-labeled AAVE Multisig, shared signers with Finding 1 |
| 5. Signer on Stani’s Safe |
30,968 |
Direct signer on Finding 2’s Safe, first ETH from same Safe |
| 6. Genesis Team infrastructure |
40,000 |
Nansen: Aave Smart Account, funded via V3 by Aave Multisig, Genesis Team signer |
| Unattributed NAY |
~25,600 |
Various addresses, no proven link to the cluster |
| Attributed total |
663,015 |
96.3% of all NAY |
Attributed to Aave founding infrastructure / Stani Kulechov wallets: 663,015 VP - 96.3% of all NAY.
Each finding uses at least two independent attribution methods: on-chain signer overlap (getOwners() calls), token transfer chain tracing, Nansen entity labels (assigned by on-chain clustering algorithms, not self-reported), ENS archival state, or funding flow analysis.
Why the Proposal Did Not Pass
The data supports a straightforward reading:
A single coordinated wallet cluster with overlapping governance infrastructure cast enough voting power to unilaterally determine the outcome.
663,015 VP against a margin of ~85,000. Remove any single finding above 85K and the proposal passes. The remaining voters, including those who voted NAY independently, did not determine the outcome.
The proposal asked delegates with material financial relationships to Aave to disclose the addresses they use to vote. It passed the temperature check. At Snapshot, it encountered concentrated opposition from addresses traceable to the infrastructure most directly affected by such disclosure requirements.
What the Data Shows
1. The NAY side was not a coalition. 96.3% of NAY VP traces to a single interconnected cluster - Nansen-labeled Aave Multisigs, Stani Kulechov’s personal Safe, and wallets funded through Aave’s own liquidation accounts. The remaining 3.7% is scattered across small independent voters.
2. The wallet architecture reduced attribution visibility. The observed patterns include: V3 supply/withdraw routing that does not present as a direct transfer on block explorers; downstream Safes with compartmentalized signer sets having zero overlap with upstream wallets; and delegation to single-purpose voting addresses with no other on-chain history. Whether interpreted as operational hygiene or concealment, the practical effect is the same: controller attribution requires forensic analysis beyond standard tools.
3. This is standing infrastructure, not a one-off. The same cluster voted as a block on the AAVE Token Alignment proposal in December 2025 - same single-purpose voting addresses, same delegation structures. If this infrastructure exists and remains undisclosed, any future governance vote can be shaped by it without the community’s knowledge.
4. The margin of defeat was narrow relative to the cluster’s resources. The proposal lost by ~85K VP. Finding 6 alone - the single-purpose wallet that sat dormant 14 months before casting its only vote - accounts for 40K of that margin. Any two of the smaller findings would have flipped the outcome. The attributed cluster (663K) exceeds the entire YAE side (603K) by 60K VP.
Closing
The on-chain record shows that 96.3% of NAY voting power on a conflict-of-interest disclosure proposal traces to one cluster’s infrastructure. The proposal lost by 85K VP. The cluster controls 663K.
Opposing a proposal is procedurally legitimate. Undisclosed concentrated control over that opposition is the governance transparency problem this proposal sought to address, and the reason it did not pass.
If any attribution in this analysis is incorrect, it can be disproven by publishing controller attestations for the cited wallets and signer sets. That is, after all, what this proposal asked for. We invite independent analysts to replicate and challenge each finding. Every address is on-chain and verifiable.
Methodology
-
Safe Transaction Service API - Safe configuration, signers, transaction history
-
Blockscout API v2 - Token transfer tracing
-
Snapshot GraphQL API - Voting records
-
Nansen Pro - Entity labels (corroborative; primary attribution relies on on-chain signer overlap and funding flows)
-
Ethereum archival state - ENS deed ownership, historical balances
-
Direct getOwners() calls - Signer set verification
Disclosure: ACI is a service provider to the Aave DAO and authored the COI Addendum proposal. This analysis uses on-chain transaction data, archival Ethereum state, Safe signer verification, and Nansen Pro entity labels. All referenced addresses are independently verifiable. Judge the evidence, not the messenger.