Twitxy Saved

Rarma

Verified account This account is verified. Learn more

@Rarma_

Building AI systems for a living. Writing about AI, politics, and whatever else holds my attention.
Joined January 2009
546 Following 7.6K Followers 34K Posts

Latest posts 5

Rarma Verified account @Rarma_
Keplr / Web3Auth social-login drain: what the chain shows so far

Two different sweepers moved funds out of the affected Keplr social-login accounts. They behave very differently. Labels below are likelihoods drawn from behaviour, not proof of who controls an address. Data to 01:03 UTC on 8 Nov. Sweep counts to 22:28 UTC on 7 Nov.

Bottom line

An attacker-likely cluster was draining before the reported pause. The earliest drain we attribute to it is 5 Nov, 01:05 UTC. That is about 44 hours before Web3Auth reported pausing its service at 21:13 UTC on 6 Nov. It drained 153 new keys after the pause, all but one before 03:00 UTC on 7 Nov. It swapped most of what it took on Osmosis. It bridged at least $783k toward Ethereum through the public Squid / Axelar route.

A separate collector held about $6.13M at 00:39 UTC on 8 Nov. It swept 45,931 keys, 86% of them after Keplr's reported warning at 04:38 UTC on 7 Nov. None of its 11 addresses had sent anything out when checked between 00:51 and 01:03 UTC on 8 Nov. Nothing had been returned to victims. Its Cosmos Hub staking sweeps carry "lsm recovery" memos. That fits the protective transfers Web3Auth has reported. It stays unconfirmed until Web3Auth publishes the recovery address list it has said is coming.

The two figures cover different keys: at least 358 drained against 45,931 swept. Do not add them, or read them as stolen versus recovered.

Some keys were drained well before the pause. 19 of the 36 keys the attacker-likely cluster drained on 5 Nov were later swept by the separate collector. The first of them was drained at 01:05 UTC on 5 Nov, about 44 hours before the reported pause.

If you used Keplr's social login

Follow only official Keplr and Web3Auth channels. Nobody legitimate will DM you to return funds. Do not sign in, connect a wallet or sign a "claim" transaction on any link you are sent. Nobody legitimate will ask for a private key, seed phrase or fee.
12 replies 6 reposts 52 likes 5.2K views View on X
Rarma Verified account @Rarma_
TERRA PROP 4853: THE LAST STAND FOR CHECKS ON THE PHOENIX FOUNDATION. VOTE NO.

4853 asks Terra to declare the dispute over a failed board vote an "emergency" and to give a supervisor the power to instruct the Phoenix Foundation (PF) board to remove a director. The failed vote is A272: an attempt on 10 October by directors Philipp Mager (@0xPhilipp, ERIS) and @MB_Aejir to remove Vini.

If Vini goes, the board becomes 2-of-3 and Philipp and MB can pass anything. That means the PF treasury, and where 31% of all staked LUNA is delegated: the stake that votes on every Terra proposal after this one.

Philipp's ERIS keys already hold upgrade control over the treasury and every TLA contract, two years after he promised to hand it over.

87% of the Yes vote is stake the PF board or ERIS placed with five validators: Philipp's own, MB's, one Philipp operates, one that pays Philipp every month, and one that is 83% PF/ERIS funded.

If it passes and Vini goes, the two directors this vote is meant to check would control the stake that checks them. This is the last vote before that happens. Vote No. Quorum is already met, so abstaining no longer stops it. Voting ends 31 Oct, 03:40 UTC.

WHAT 4853 DOES
Philipp submitted it from his ERIS validator key. It asks Terra governance to:
- declare "the persistent Stewardship dysfunction surrounding Proposal A272" an Adverse Event under PF's bylaws
- appoint Martin as Emergency Supervisor ("MB and I support Martin", Philipp, forum post 85)
- confirm his power under Art. 2.4.5.4 to instruct directors "to facilitate removal"
- pay him up to $10,000
His declaration of independence comes after the vote, with his acceptance. Not before.

While an Emergency Supervisor is in place, Bylaws Art. 1.10 says the Veto Multisig's veto "may be disregarded". That check on the board switches off.

The proposal says it does "not predetermine findings" or "directly remove any Director". On 25 Oct, in Lion DAO's public Telegram while Lion's members were voting on 4853, Philipp wrote: "any director will be legally liable if he doesn't follow the notice of removal." Asked whether Vini's seat would be filled if he is dismissed: "Yes." On what happens if the board doesn't follow the supervisor: "it will go to governance and sudo the action. And afterwards the supervisor is contacted for legal action."

What follows a removal is not in doubt. A272's own on-chain messages "remove Vini from the multi-sig and convert it to a 2-out-of-3 multi-sig". Philipp, post 5: "a removal necessarily will always come with a move to 2-of-3". Asked to make it binding that no removal leaves him and MB as sufficient authority alone, he replied (post 57): "I couldn't care less about such a clause." He offered instead an off-chain promise to vote only No or Abstain during the 2-of-3 period, except to onboard new directors. He says new directors would be added later, as a separate step (post 57). Neither the promise nor the new directors are in the proposal you are voting on.

THE "EMERGENCY" IS A FAILED VOTE
A272 tried to remove Vini. Philipp and MB voted Yes, Rose voted No, Vini did not vote. It needed 3 of 4 and failed on 10 Oct. That is the "Adverse Event".

Rose, a director (post 72): "I do not consider the failure of A272, or a disagreement between directors, to constitute an Adverse Event in itself. A failed vote is a legitimate outcome of a governance process." She says any independent review should be able to examine "any of the four directors, including my own".

Since A272 failed, the board has executed 14 proposals. Vini voted Yes to pay Philipp, MB and Rose for October (A277 to A279). There is no October stewardship payment for Vini. In Lion DAO's Telegram, Philipp listed what "will now fail", including "You want to pay vini -> fail." Paying Vini needs three of four votes and he has voted to pay the other three, so it only fails if two of them refuse.

PF already has a standing Supervisor under its bylaws. A272 was never put to them. Philipp's reason (post 91): "not required, and doesn't have the authority to instruct removal."

WHO IS VOTING YES
As of 25 Oct, 10:06 UTC: Yes 48.0M LUNA, No 25.1M, No with Veto 1.6M, Abstain 18.9M. Yes is 64.3% of decisive votes. Turnout crossed the 30% quorum at 06:50 UTC on 25 Oct, with HighStakes' Yes.

Five validators cast 98.9% of the Yes:
- Galactic Punks, 14.3M. 89.4% of its stake was placed by PF/TLA or ERIS's hub. Its operator key pays Philipp's validator key every month: 116,735 LUNA in ten payments since February.
- ERIS, 8.6M. Philipp's own validator. 98.9% PF/ERIS stake.
- Orbital Command, 8.3M. MB's validator ("MB aejir (OC/PD)" in the Phoenix Directive's own 2024 proposal). 73.5% PF/ERIS stake.
- Lion DAO, 8.3M. Run by Philipp, per Lion's founder and a Lion team member. 94.7% PF/TLA stake. Its operator key also pays Philipp's key every month: 105,227 LUNA in ten payments, each sent within five minutes of GP's.
- HighStakes, 8.1M. 83.2% PF/ERIS stake.

Stake placed by the PF board or ERIS's hub is 87% of all Yes voting power. The validators run by Philipp and MB, the two directors who voted to remove Vini (ERIS, Orbital Command and Lion), are 52% of it on their own.

Across the network, PF/TLA places 30.9% of all staked LUNA and ERIS's hub another 10.5%. 18 of the top 25 validators get more than half their stake from them.

Dependence is not destiny. Smart Stake gets 68.9% of its stake from PF/ERIS and voted No. That is what independence looks like.

The Lion and GP payments may be fees for running those validators. Either way, two of the validators voting Yes on Philipp's proposal pay Philipp every month.

LION DAO
In a pinned post in Lion DAO's public Telegram (12 Jan 2024), Lion's founder Ryan Lion wrote that Philipp "operates our validator", and that on the Terra delegation committee they both "abstained from any vote relating to delegations to Lion DAO". That committee's report lists Philipp under "Eris + validator operations (The Lion Dao)". On the PF board since, Philipp has voted Yes on all ten proposals setting the delegations Lion receives (A4, A5, A6, A20, A21, A194 to A198).

On 25 Oct, a Lion DAO team member Usaf, confirmed it in the same group: "Phil provides hosting services the same way Hexagon do but is still required to abide by DAODAO signalling vote."

On 4853, Lion's validator first voted Abstain (24 Oct, 04:18 UTC). Lion DAO then opened that signalling vote (A33), closing 26 Oct at 13:16 UTC. Four wallets voted. Ryan's own wallet ("RyanLion" on DAO DAO) is 99.9% of the Yes weight. ERIS's ampROAR hub, which votes 58% of Lion DAO's staked ROAR, abstained, as it has on every Lion proposal it has voted on. No one voted No. As it stands, Lion's members' vote comes down to its founder.

The validator (Philip) did not wait for it. It switched to Yes at 05:06 UTC on 25 Oct, 13.5 hours after Ryan voted and 32 hours before the vote it is "required to abide by" closes. Less than six minutes later, the same key executed the ERIS multisig proposal that handed admin of ERIS's ampLUNA hub to the PF board.

Ryan also sits on PF's Veto Multisig: the check that Art. 1.10 lets the board disregard once an Emergency Supervisor is appointed. So do Philipp, who proposed 4853, and Rebel, who argued for an Emergency Supervisor on the forum (post 43) and is listed as a partner in Orbital Command in Terra's 2022 governance records. Three of the six members of the check that 4853 switches off are on its side.

While Lion's members were voting, Philipp was in their chat. On whether the supervisor's advice binds the board: "It is legally binding, however when they will not follow it, it will go to governance and sudo the action. And afterwards the supervisor is contacted for legal action." Six minutes later: "The only vote is about getting an independent review." Four minutes after that: "any director will be legally liable if he doesn't follow the notice of removal."

THE KEYS HE NEVER HANDED OVER
When Philipp launched the Terra Liquidity Alliance in August 2024, his proposal said: "Contract Admin will be transferred to PD Stewardship after 1 month" and "Ownership transferred to PD Stewardship after 3 months."

Today, the upgrade admin of the PF Treasury and all 13 TLA contracts (the registry and the 12 contracts it lists) is a single ERIS key, terra1q0vny...q5tew. The TLA registry, which tells every TLA contract who controls it, is owned by another single ERIS key, terra1kefa...klgzl. Whoever holds a contract's upgrade admin can replace its code. It's Philip.

Philipp (post 55): "Given my unresolved concerns about the current Board, ERIS Team has not considered transferring that authority appropriate."

PAY
Every director votes on their own pay. On board proposals paying their own wallets, Philipp voted Yes 17 of 19 times, MB 10 of 11, Vini 12 of 16, Rose 6 of 11.

October pay: Philipp 8,640 USDC, MB 5,640, Rose 4,800.

On top of that, PF approved grants to Philipp's validator key for ERIS projects he proposed: Votion $50,000, Creda $210,000, Estima $125,000. He sat out the final Creda and Estima approvals, then voted Yes on every milestone release paying his own key: $305,000 approved for payout so far. Votion also received $50,000 from the Cosmos Hub's AADAO, whose oversight report lists "Philipp Mager" as its grantee representative.

Add the monthly LUNA from Lion and GP: 221,962 LUNA since February.

Vini isn't clean either. He proposed and voted Yes on A249, which paid 16,900 USDC to his own wallet for a security audit. He voted Yes on A159, which raised his own monthly pay from 2,800 to 5,000 USDC. Those are real, and an investigation should cover them.

That is the point. Self-dealing on this board is not one director's problem. It is all four. 4853 builds its case around one of them. Asked to put directors' votes on payments to themselves and the conflict rules A272 relies on into scope (posts 37 and 39), Philipp said "No" to "an open-ended investigation of every Director, every grant and every PF decision ever made" (post 38), while saying he has "no issue" with his own actions being reviewed. MB (post 40): "No. The scope of this prop is not going to be changed." The final text lets the supervisor examine "the conduct of any Director where relevant". It does not name self-payments, conflicts or retained contract control.

THE FORUM
The four directors are the forum's admins.
- 19 Oct: two critics of 4853 were suspended until 2 Nov for "No constructive purpose to their actions other than creating dissent within the community". Lifted within half an hour.
- 20 Oct: one of them was silenced until 30 Nov, a month after this vote ends, for "Personal attacks, swearing, repetitive and obnoxious posts". One of his posts did swear.
- At least three posts were removed by staff, not by their authors. Two were from Free Do's validator operator account, which voted No with Veto: one called for term limits, a fifth director and removing both Vini and Philipp; the other linked to a site documenting Philipp's payments. The third, from the other suspended critic, asked how many times the proposal's author had voted on funding his own projects. The on-chain answer: at least seven (A107, A116, A175, A209, A213, A215, A271).
Only PF's staff action log shows which admin did what. PF can publish it.

PF's own X account (@phoenix_dir) used its 25 Oct post to promote a different Terra governance proposal. Not this one.

THE LAST STAND
If 4853 passes:
- a failed board vote becomes an "emergency", and a precedent
- the Veto Multisig can be disregarded
- a supervisor can instruct the removal Philipp and MB could not win on 10 Oct
- the board becomes 2-of-3, and Philipp and MB are the 2
- they control the treasury and where 31% of staked LUNA sits, while Philipp's ERIS keys keep upgrade control of the treasury and the TLA

Any later Terra vote to undo it would be cast largely by validators whose stake they place. Look at who is voting Yes today.

What a fair process looks like: one standard for all four directors, with payments to themselves, conflicts and retained contract control written into scope. Contract admin handed over as promised in 2024. A binding rule that no removal leaves two directors with sufficient authority alone. A supervisor's independence declared before the vote, not after. None of that is in 4853.

Vote No, or No with Veto. Quorum is met, so Abstain no longer blocks it. It flips if GP moves to No, or any two of the other four Yes validators do, or with about 21.3M LUNA of new No votes. If you delegate to a Yes validator, vote yourself: your vote overrides theirs for your stake. Voting ends 31 Oct, 03:40 UTC.

SOURCES
Proposal 4853: mintscan.io/terra/proposal…
Forum thread (posts 5, 37 to 40, 43, 55, 57, 72, 85, 91): forum.phoenix.money/t/phoenix-foun…
PF Bylaws (Art. 1.10, 2.4.5.4, 2.6): github.com/phoenix-direct…
A272: daodao.zone/dao/terra1k8ug…
PF board proposals cited (A4-A6, A20, A21, A107, A116, A159, A175, A190, A194-A198, A209, A213, A215, A249, A266, A271, A277-A279): daodao.zone/dao/terra1k8ug…
Philipp's full name: - AADAO oversight report (Votion, "Grantee Representative: Philipp Mager"): forum.cosmos.network/t/aadao-wind-d… - GitHub commit record, Terra Core, 24 Jul 2024 (author name "Philipp mager", account 0xphilipp; the same author name is on his ERIS commits from 2022 to 2024): api.github.com/repos/terra-mo…
TLA launch pledge, Aug 2024: github.com/phoenix-direct…
PF Treasury contract: chainsco.pe/terra2/address…
TLA registry: chainsco.pe/terra2/address…
ERIS multisig handover, 25 Oct: mintscan.io/terra/tx/86E59…
Lion to Philipp's key, 19 Oct: mintscan.io/terra/tx/B77EA…
GP to Philipp's key, 19 Oct: mintscan.io/terra/tx/C940E…
Phoenix Directive 2024 proposal (MB "OC/PD", Philipp's key): github.com/phoenix-direct…
Delegation committee report (Philipp, Lion validator operations): github.com/phoenix-direct…
Lion DAO members' vote A33: daodao.zone/dao/terra1tker…
Lion validator Abstain: mintscan.io/terra/tx/95313…
Lion validator Yes: mintscan.io/terra/tx/4B356…
Ryan's Veto Multisig wallet to the RyanLion wallet: mintscan.io/terra/tx/1D67E…
PF Veto Multisig: daodao.zone/dao/terra15l75…
PF doc listing the Veto Multisig members' wallets (Philipp, Rebel, Ryan): github.com/phoenix-direct…
Rebel Defi, "Partner - Orbital Command" (2022): github.com/phoenix-direct…
@phoenix_dir, 25 Oct: x.com/phoenix_dir/st…
Lion DAO public Telegram (t.me/Lion_DAO): Ryan Lion, 12 Jan 2024 (pinned).

$LUNA
Terra Phoenix Verified account @phoenix_dir Oct 25
A new Terra Governance proposal outlines the recovery process following the legacy Terra Station Desktop Ill Bloom compromise.

The proposal includes potential reimbursement of verified claims from the PF Treasury.

🌐 forum.phoenix.money/t/update-terra…
View quoted post on X
8 replies 6 reposts 53 likes 9.6K views View on X
Rarma Verified account @Rarma_
I think now is a good time to provide an update on my Cosmos 10k Challenge.

Disappointed I'm down a bit from last time. But overall I think the investments have worked out well.

23 Oct 2026 vs 26 Aug 2023.

cosmos:native
$OSMO
secret:native
$LUNA $JUNO
$BTC
$ETH
Rarma Verified account @Rarma_ Aug 14, 2025
We’re back! Another update on the famous Cosmos 10k challenge!

Significant improvement on last time. And very happy with my original 10k investment.

I started this one 28/08/2023. And as you can see my investments are significantly up.

Let’s see where we go from here. But I have very high confidence.

Now vs 2023

$ATOM $OSMO $SCRT $LUNA $JUNO $BTC $ETH
View quoted post on X