Rocket Pool
Tier 1Security Analysis Report
TVL
$1.11B
Sections
7
report sections
Citations
516
evidence references
Version
v4
10/29/2026
Moderate Risk
Safety Score · out of 10
Executive Summary
Executive Summary: Rocket Pool
Rocket Pool is a mature Ethereum liquid staking protocol (about USD 1.12bn of value locked, all on Ethereum) with a named team and a long audit history. The cross-analysis rates the composite risk as moderate, with a few named combined risks. Direct rETH exits go through a very small buffer, the newest upgrade (Saturn 1) has not been tied to its audits, trusted committees control the rETH rate and contract upgrades, and several live settings were not read on-chain. Evidence is strong on how the protocol works and weaker on live admin settings, so the governance picture is incomplete rather than proven bad.
Safety Rating
Overall Score: 5.0/10 Dimension scores: Code security 4.3 · Economic security 5.3 · Team transparency 5.7 · Governance 3.5 · Track record 6.7
- Code security
- 4.3 points
- Economic security
- 5.3 points
- Team transparency
- 5.7 points
- Governance
- 3.5 points
- Track record
- 6.7 points
Governance has a low score mainly because key facts were not read on-chain (upgrade delay, Security Council signers, guardian powers, voter concentration). The cross-analysis says the governance score reflects missing data more than a measured weakness . The cross-analysis puts overall confidence at moderate .
Scores by Dimension
| Dimension | Key Finding |
|---|---|
| Code Security | Four independent firms reviewed the Saturn upgrade, but live code was never compared with any audited version. |
| Economic Security | rETH earns staking rewards and operators take first losses, but the rate comes from a ten-member oracle committee and direct-exit liquidity is under 1% of supply. |
| Team Transparency | The founder, general manager and company are named; the company's jurisdiction and treasury reporting are not in the sources. |
| Governance | Two on-chain DAOs plus a Security Council veto, but the guardian is a single key and council signers and voter concentration are unconfirmed. |
| Track Record | Live on Ethereum since 2021-11-09 with several upgrades delivered; one recorded 2023 phishing loss of about USD 3.8M, whose victim no first-party source confirms. |
What This Protocol Does
Rocket Pool lets anyone stake ETH through a decentralised network of node operators and receive rETH, a liquid receipt token that grows in value against ETH. The products in scope are the rETH token, the RPL governance and collateral token, the node-operator system (minipools and the newer megapools) and the two DAOs that govern it. DefiLlama records TVL on Ethereum only; no inactive chains were reported in the worker reports.
Key Strengths
- Review depth on the Saturn 1 upgrade. Hashlock (January 2026), Bailsec, Cantina and Sigma Prime reviewed it. Hashlock's verified scope covers megapool, deposit, node, token and network contract paths, and its one Medium finding is marked resolved . The audit index could not confirm the scope of the Sigma Prime Saturn review or the Bailsec report . Several of those copies are project-hosted with unverified provenance, which lowers confidence but not the credit.
- Fixed findings. Of 458 distinct audit findings, 235 are marked fixed, including 9 of 11 critical groups.
- Operators take losses first (megapool validators). Operators bond 4 ETH per validator and absorb losses up to that amount.
- Organic rETH yield. rETH returns come from Beacon Chain, fee and MEV rewards, not token emissions.
- Public, named team and open governance process. Every change starts as a public proposal. The official process page lists a one-week upgrade delay and a Security Council veto with a 33% quorum . The live
upgrade.delaywas not read on-chain, and the team's January 2026 statement gives a 2-week minimum, so the delay is not confirmed .
Key Risks
- rETH exit discount under stress (rETH holders). Magnitude: unknown, because DEX depth was not measured. Likelihood: medium (assessment). The rETH contract and deposit pool hold about 2,646 ETH of collateral, a collateral rate of 0.73%, against supply worth about 362,000 ETH. If sellers outnumber DEX liquidity, rETH could trade below its protocol rate. A community forum post (source reliability: community publication) says mainnet rETH liquidity fell from over USD 30M to about USD 2.5M; this is an unconfirmed lead. A captured Balancer page (source reliability: unverified publisher) says most pools move to withdrawals only on 30 October 2026, which could close a main exit venue; this is also an unconfirmed lead. In an exploit scare the Security Council can act instantly only with pause-type powers, and the sources describe no global pause.
- Loss of staker ETH through Saturn 1 bugs, or a loss above the 4 ETH bond (rETH, megapool operators). Magnitude: severe in the worst case, not sized in USD. Likelihood: low to medium. Bailsec cannot state with high confidence that no critical issues remain. The docs do not say who absorbs a loss above the bond. A separate conditional path is megapool exit accounting: if the node operator fails to notify an exit and the Oracle DAO fails to challenge in time,
distributetreats principal as rewards. The team classes this as intended behavior and relies on the challenge process, which also fines the operator, and no realised shortfall is recorded. No insurance or reserve fund was found, and the Immunefi bounty maximum is USD 150,000. - Trusted-key control of the rETH rate and of upgrades, plus adjustable RPL rewards (rETH, RPL stakers, node operators). Magnitude: about 7,200 ETH per 2% rate error; a malicious upgrade could in principle reach the whole TVL. Likelihood: low for key-control events. In practice 6 of 10 oracle members must agree on the rate. An upgrade can replace any registered contract, including those holding ETH. The live rate limit was not read on-chain. RPL staker rewards depend on adjustable splits and on inflation planned to phase out with Saturn 2. Assessment: a change to RPL staker rewards is likely over time, because the documented plan phases out the inflation rewards . The exposed reward stream is smaller than total issuance and its size is a gap .
- Unmatched audits and unfixed older findings. The 52 critical, high or medium findings not marked fixed include two 2021 critical groups on deployment-phase powers. The audit coverage table shows deployed code was not compared with the audited version; this is "not compared", not a measured mismatch.
- Single-key guardian. The guardian address read from
RocketStoragehas no contract code, so it is a single key with zero-delaysetGuardian. Its other remaining powers are unresolved. A related model observation is published only as a hypothesis, informational.
Critical Findings
- No confirmed critical exploit is reported. Two 2021 ConsenSys Diligence critical findings are not marked fixed: DAO takeover during deployment and open settings until the contract is declared deployed . The audited code revision was not available to check them against code. The on-chain registry now reports the deployment phase as finished, which closes the settings window only . Whether the DAO-takeover issue still matters depends on guardian powers that were not traced .
- The Bailsec review of Saturn reports 144 findings including 26 High, and a follow-up commit still had fund-theft and DoS issues. These are audit findings, not confirmed live exploits.
- The team said in January 2026 that the Security Council's only member is a multisig of three team members. The current owners behind the council contract are not identified on-chain, so that composition is not confirmed.
Documented vs On-Chain
- The oracle committee: Exponential says consensus needs 9 of 16 members. On-chain at Ethereum block 26294960 (2026-10-29) the committee has 10 members and a vote quorum of 5.1.
- Redemption buffer: the Saturn 1 docs describe a 1% buffer and Sigma Prime refers to a 10% buffer; on-chain at block 26294960 the collateral rate is 0.73%.
- Upgrade delay: governance cites a one-week delay and the team's January 2026 "2 weeks" minimum; the live value was not read on-chain and is not confirmed.
- Security Council: the team (January 2026) says the member is a multisig of three team members; the on-chain read of a contract with threshold 2 and three owners does not show whether those owners are the team members.
Recommendation
Rocket Pool suits users who accept staking-protocol risk and can hold rETH through a possible short-term discount. Its yield is organic, its team is public, and the Saturn 1 code had four independent reviews. Large holders should size positions against the thin direct-exit buffer. Holders beyond the 0.73% buffer would need DEX liquidity, whose depth was not measured, or validator exits. No check of the audits or governance settings would establish a fast exit. A measured 2% price-impact depth across venues would help assess the exit-discount risk , but no maximum delay is stated for validator exits . Separately, users who want proof that live contracts match audits, or confidence in the upgrade controls, should wait for a deployed-code comparison, a read of the live upgrade delay and rate limit, and identification of the Security Council owners.
Report Sections
- Protocol Overview
- Code Security
- Economic Security
- Social & Team
- Governance
- Cross-Risk Analysis
52 citations · click markers to view evidence
Protocol Overview
Protocol Overview: Rocket Pool
What It Does
Rocket Pool lets anyone stake ETH through a decentralised network of node operators and receive a liquid receipt token, rETH, that grows in value against ETH. Staking means locking ETH to help run Ethereum's validators, the programs that check blocks and earn rewards. A user can stake from 0.01 ETH, and a node operator can run a validator with a fraction of the usual 32 ETH. The protocol lives only on Ethereum mainnet for TVL purposes; DeFiLlama records about $1.12bn of TVL there (snapshot 2026-10-29). The evidence is strong for the mechanics (official docs plus on-chain reads) and weaker for how safe the newest upgrade is, because its audits are recent and one reports unresolved issues.
- Category: Liquid staking (DeFiLlama category "Liquid Staking") .
- Products and tokens in scope: the rETH liquid staking token, the RPL governance and collateral token, the node-operator system (minipools and the newer megapools), and the two DAOs that govern it. The docs name "two tokens. RPL and rETH" .
- Who uses it: liquid stakers who want ETH staking rewards without running hardware, and node operators who want higher returns on less capital. The docs describe two groups: tokenised staking from as little as 0.01 ETH, and operators who run a node .
- Node operator economics: "Unlike solo stakers, who are required to put 32 ETH up for deposit to create a new validator, Rocket Pool nodes only need to deposit 4 ETH per validator" . The other 28 ETH comes from liquid stakers' deposits .
How It Works
Liquid stakers and node operators meet in a shared ETH pool. The stakers supply most of the capital and the operators supply the machines, a small bond and the risk of first loss.
- A staker deposits ETH. The staker sends ETH to the deposit pool contract (
RocketDepositPool) and receives rETH from the token contract (RocketTokenRETH). The docs say the staker's "role is to deposit ETH into the deposit pool which will enable a node operator to create a new Beacon Chain validator" . - A node operator bonds ETH and runs a validator. The operator registers a node and puts up a 4 ETH bond per megapool validator. A megapool is one contract per node that groups its validators, replacing one contract per validator . The pool's 28 ETH is matched to the bond to form a full validator.
- The validator earns rewards. To the Ethereum Beacon Chain a megapool validator "looks exactly the same as a normal validator" . Rewards are consensus-layer rewards, priority fees and MEV (extra tips paid for block ordering).
- Trusted oracle nodes report back to Ethereum. Smart contracts cannot see the Beacon Chain, so oracle nodes run by trusted community members relay "total staking rewards and validator exit status" . The rETH rate then moves: "The rETH/ETH exchange rate is updated approximately every 24 hours based on the Beacon Chain rewards earned by Rocket Pool node operators."
- A staker redeems. The staker can swap rETH back to ETH at the contracts, but only when there is ETH on hand: "Trading rETH back for ETH directly with Rocket Pool is only possible when the staking pool has enough ETH in it to handle your trade." . Otherwise the staker sells rETH on a DEX, usually at a small premium cost .
- Losses fall on the operator first. If a validator exits with less than 32 ETH, the operator absorbs the loss: "if you exit with a balance of 31 ETH, then your megapool validator lost 1 ETH from its initial 32 ETH deposit. You will receive 3 ETH, and 28 ETH will be returned to the staking pool."
- (actor)
- (contract)
- (token)
- (actor)
- (contract)
- (role)
- step 1flow
- step 1flow
- step 2flow
- step 4dependency
- step 5flow
Core mechanism. Rocket Pool is a pooled-staking system with operator bonds. Liquid stakers hold rETH, a token whose value rises against ETH as validator rewards accrue. Node operators put up first-loss capital and keep a commission.
Version history that matters. Minipools (one contract per validator) required a larger operator bond. Saturn 0, introduced on 28 October 2024, "removed the mandatory RPL bond for new minipools and shifted RPL to earn ETH via a commission share" . Saturn 1 added megapools, a 4 ETH bond, standard and express queues and adjustable commission controls . That feature page still uses future tense. The current node docs and the active-contract list describe megapools in present tense, and the megapool factory contract has deployed code on Ethereum at block 26296361 . Assessment: Saturn 1 is live on mainnet; the project's timeline page lists its mainnet launch as 18 February 2026 .
Key contracts and roles (Ethereum mainnet, from official docs).
| Contract | Role |
|---|---|
RocketStorage (0x1d8f…0fa46) | Registry and data store. "the Storage contract is used to find all the other deployed contract addresses as needed" |
RocketTokenRETH (0xae78…6393) | The rETH token and its exchange rate |
RocketTokenRPL (0xD335…1a51f) | The RPL token |
RocketDepositPool | Holds staker ETH awaiting assignment to validators (address caveat below) |
RocketMegapoolFactory (0xD5bf…E38b) and RocketMegapoolDelegate | Create and run per-node megapools |
RocketNetworkPrices (0x7518…fc37) | Stores the RPL price that oracle nodes submit |
RocketDAONodeTrusted* and RocketDAOProtocol* | The oracle DAO and protocol DAO logic (see Technical Architecture) |
Address check: the docs table lists RocketDepositPool at 0xCE15…4fdC, while the on-chain file read a Deposit contract at 0xdd3f…5af8 (version 3). A direct RocketStorage.getAddress lookup for the rocketDepositPool key at block 26296422 returned 0xce15294273cfb9d9b628f4d61636623decdf4fdc . The live deposit pool is therefore 0xCE15…4fdC. The status of 0xdd3f…5af8 was not looked up; the registry lookup does not say what that contract is.
Yield Sources
rETH holders earn Ethereum staking rewards, not token emissions. Node operators earn a commission plus optional RPL incentives, and part of the operator reward is paid from RPL inflation.
- Staker yield. Value comes from the Beacon Chain, priority fees and MEV: "The ratio includes rewards that Rocket Pool node operators earn from:" followed by "The Beacon Chain itself" , "Priority fees from block proposals" and "MEV rewards from block proposals" . The rate only goes up against ETH as rewards accrue, apart from losses (item 6 above and Key Risks).
- Sustainability. The staker yield is organic: it is paid by the Ethereum network. The docs give a May 2026 network figure: "As of 5/2026, overall APR is ~2.84%, with 2.63% being consensus layer APR, and 0.22% being execution layer APR." . This is the validator-level rate before Rocket Pool commission. The sources give no net rETH APR history.
- Current rETH rate. On-chain at Ethereum block 26294960, the rETH exchange rate was 1.1746 ETH per rETH and total supply was about 308,566 rETH .
- Operator commission. "Each megapool validator earns 5% commission on rewards generated from the 28 protocol ETH." . The docs say this is 35% more than solo staking. Minipool commission is documented inconsistently. The glossary says "New minipools activated are now at a set 14% commission rate." The German node-staking documentation says minipools created after 28 October 2024 had a temporary bonus commission of up to 14% via the Smoothing Pool, that the bonus ended four reward intervals after Saturn 1 (the last paid in June 2026), and that these minipools now receive a 5% contract commission . The glossary figure is therefore treated as superseded for post-October 2024 minipools; the rate for other minipool arrangements is not stated in the sources.
- RPL incentives. Operators who stake RPL get a share of protocol commission paid in ETH and RPL issuance rewards. "As of Saturn 1, vote-eligible staked RPL will get a share of total protocol revenue" . The docs say operators can "earn max yield on RPL up to 15% of the value of your total borrowed ETH" .
- Inflation dependence. The majority of RPL inflation goes to operators as rewards, and "these rewards are planned to be phase out with Saturn 2" . The RPL contract shows a daily inflation interval rate of 1.00013368 , which compounds to roughly 5% a year (a calculation, not a quoted figure). Operator RPL income is therefore token-emission funded today, and the docs plan to phase it out.
- Revenue split. The DAO can change how protocol ETH revenue is divided between RPL stakers, operators and rETH holders: "The split of the protocol's ETH revenue between RPL revenue, node operator commission, and rETH rewards can be changed dynamically by the DAO" . A change could lower the share that reaches rETH holders.
Dependencies & Integrations
The protocol depends on Ethereum itself, on a small trusted oracle committee, and on outside venues for rETH liquidity and pricing.
- Ethereum Beacon Chain. Validators run there. Deposits, exits and slashing are Ethereum rules, not Rocket Pool rules.
- Oracle DAO (oDAO). The contracts rely on it for Beacon Chain facts and prices. The on-chain read shows 10 members, a minimum of 3 required and bootstrap mode disabled .
- Market and NAV price feeds listed by the docs. NAV is the protocol's own exchange rate; the market feeds are third-party. The docs list "RP oDAO Rate (Mainnet)" as the NAV oracle, plus Chainlink on Ethereum and Arbitrum and a Maker feed for rETH/USD . Rate providers for Optimism, Polygon and zkSync Era come from Rocket Pool's own repositories .
- Maker rETH/USD feed freshness. The on-chain read of the Maker feed shows a last-update timestamp of 1742497200 (about 20 March 2025) with an hourly interval . Assessment: this feed looks dormant at the 2026-10-29 read, so integrators should not rely on it.
- DEX venues. Rocket Pool recommends a Balancer composable stable pool because it honours "the true exchange rate reported by the Oracle DAO", and lists Uniswap as an alternative . A depth check of DEX liquidity returned no candidates, so DEX depth is not measured .
- Downstream DeFi. Docs list rETH in vaults and lending-style integrations (Aura Finance, Gamma, Notional Finance, Contango, Flat Money) .
- Other chains. rETH and RPL also exist on Arbitrum, Optimism, Polygon, Base and zkSync Era . The docs say the L2 path means "bridging your existing ETH (or other tokens of choice) onto an Ethereum Layer 2 network such as" an L2, then swapping . The bridge used for each L2 token was not assessed. DeFiLlama attributes no TVL to these chains.
TVL & Adoption
DeFiLlama's latest captured snapshot (collected 2026-10-29) shows about $1.118bn of TVL, all on Ethereum.
- Total: "tvl: 1118047553.4013996" , collected 2026-10-29T05:08 UTC.
- By chain: "Ethereum: 1118047553.4013996" . No other chain is listed.
- Staking line: DeFiLlama also reports a separate staking figure of about $11.4m, listed as "Ethereum-staking"; the sources do not state which token it measures .
- Recent movement: the same snapshot shows a 7-day change of -11.7% . Part of that may be ETH price; the sources do not split price from flows.
- Scale on-chain: about 308,566 rETH at 1.1746 ETH each is roughly 362,000 ETH (calculation from the on-chain figures above).
- History: no dated historical TVL series is in the sources, so no trend is given.
- Ethereum
- 1.118 USD bn
- Ethereum-staking line
- 0.011 USD bn
- Integrations: see the list in Dependencies. Adoption beyond TVL, such as share of Ethereum stake or operator counts, is not in the sources.
- Past incident record: DeFiLlama's hack database lists one record for Rocket Pool: 30 March 2023, $3.8m, classed as phishing / social engineering . It is a single database record and the sources give no further detail.
Technical Architecture
Rocket Pool is governed by two on-chain DAOs plus a guardian key, and its contracts are replaced through a registry rather than proxies. The guardian is a single externally owned account on Ethereum at the block read.
Upgrade pattern. All contracts look each other up in RocketStorage, and writes to storage are limited to registered network contracts. The on-chain privileges table shows setAddress guarded by onlyLatestRocketNetworkContract . The Ethereum contracts in the read show no standard proxy slot. Upgrades therefore swap contract addresses in storage. The upgrade functions are upgrade and bootstrapUpgrade on RocketDAONodeTrustedUpgrade, and proposalUpgrade on RocketDAONodeTrustedProposals . The delay for these is unknown (null), not zero.
Who controls what (Ethereum, block 26294960).
| Actor | Power | Evidence |
|---|---|---|
Guardian 0x0ccf…df21e | Can call setGuardian and setDeployedStatus on RocketStorage with no delay. It has no deployed code, so it is a single key (EOA) | ; |
| Oracle DAO (oDAO) | 10 members, 3 minimum. Submits balances and prices; its proposals can change settings and upgrade contracts | |
| Protocol DAO (pDAO) | RPL holders; bootstrap mode is disabled | ; |
| Security Council | Elected by the pDAO; can propose and execute changes without delay; at Houston launch the pDAO guardian was its only member; on-chain at block 26296425 the security contract reports 1 member, which is not the guardian | ; ; ; |
| GMC and IMC multisigs | Safe wallets with 7 owners and a 4-signature threshold each | ; |
The Security Council row paraphrases a Japanese-language docs page. The registry points the security contract to 0x84ae6d61df5c6ba7196b5c76bcb112b8a689ad37 . The single member, 0x6c565af34f15de064b56afedf7b0f59c15bb20fe, has contract code of 171 bytes , so it is not a single key. It returns a threshold of 2 and three owners , which are 0xac69…b09e, 0x22a3…c4b3 and 0xfdcc…f538. The owners were not checked for being keys or contracts, and the sources do not say who they are. The threshold and owners come from these calls only; no source describes the contract. The GMC and IMC names appear in the docs under "Management Committee Wallets" . What each committee controls is not stated in the sources.
Parameters set by governance (read on-chain). The oDAO proposal settings were read as: vote delay 604800 s (7 days), vote time 1209600 s (14 days), action window 2419200 s (28 days) and cooldown 172800 s (2 days) . Other oDAO-set values include a 12-hour minipool scrub period and a 51% quorum . Among Saturn 1 settings, the Japanese-language Saturn 1 page says the bond requirement is a pDAO-adjustable parameter .
Upgrade safeguards. The docs say "A mandatory delay period will be enforced before protocol upgrades go live, during which the security council can veto changes." . The oDAO also gets limits: "Saturn 1 adds guardrails for the oDAO to cap how much can be penalized over time" and constraints on balance data to protect "the rETH protocol rate from manipulation or faulty reporting" .
Emergency mechanisms. The governance page says the Security Council can act in a live incident "with a strictly whitelisted set of pause-type powers" , and the same page gives recent examples: "disabling solo migration, node deposits, & node registrations" . The sources give no complete list of these powers and do not describe a global pause. They also describe the Security Council's undelayed changes and veto right, and the guardian's setDeployedStatus. Insufficient data for a full list of emergency controls. The control read examined the Deposit contract at 0xdd3f…5af8, not the 0xCE15…4fdC address the registry returned as the live deposit pool. On that examined contract, the fund-moving functions withdrawExcessBalance and nodeCreditWithdrawal carry the guards onlyThisLatestContract + onlyLatestContract, which the parser did not resolve, and their holders were not resolved . Whether these readings apply to the current deposit pool is not established.
Security reviews of the current system. The docs list audit reports by Sigma Prime, Consensys Diligence, ChainSafe, Cantina and Bailsec, among others. The Saturn 1 (megapool) round is the most relevant to funds today. The Cantina, Bailsec and Sigma Prime Saturn reports are project-hosted copies attributed to those auditors: the audit index notes they were discovered via DeFiLlama and that the publication host is not verified as auditor-controlled . Their findings and fix statuses below are as reported in those documents. Hashlock's report is hosted on the auditor's own site .
- Cantina (report dated 13 January 2026): "63 total issues (2 Critical, 4 High, 3 Medium, 43 Low, 1 Gas Optimization, 10 Informational), with 46 fixed and 17 acknowledged" and the visible list is partial . Scope includes megapool, deposit pool and DAO contracts.
- Sigma Prime Saturn review (February 2026): "The review identified 19 issues (1 High, 4 Medium, 7 Low, 7 Informational)." . It was time-boxed and limited to one code diff (6f07f2a..269782f) , and its code scope was not confirmed against the audit index.
- Hashlock (January 2026): one Medium finding, resolved .
- Bailsec (January 2026): "144 findings in total, including 26 High", with fixes for several refactored contracts "in open status pending re-review"; the report states that the auditors "cannot state with high confidence that no critical issues remain" . Its scope review did not agree with the audit index, so its coverage of the deployed code is unconfirmed. Whether later fixes were made or re-reviewed is not stated.
- Older audits (2021 to 2024) cover earlier versions: pre-Atlas minipool code, Houston DAO changes (Consensys Diligence, Sigma Prime, ChainSafe April 2024). They do not cover megapools.
- Bug bounty: Immunefi, with critical rewards of "10% of the funds directly affected up to a maximum of" USD 150,000 .
- (contract)
- (role)
- (governance)
- (governance)
- (contract)
- (role)
- (external)
- control
- control
- control
- control
Key Risks (Protocol-Level)
User ETH sits in the deposit pool, the rETH contract and Ethereum validators run by third-party operators, so the main risks are validator losses, contract bugs in the new megapool code and the trusted oracle committee.
- Where funds sit. Staker ETH is in the deposit pool and rETH contract until assigned, then in Beacon Chain validators. On-chain at block 26294960, the rETH contract reported total redemption collateral of about 2,646 ETH, a collateral rate of about 0.73% of rETH value . This is a total across the contracts, not necessarily ETH held in the rETH contract alone: an earlier version of the redemption code (2021 audit excerpt) pulls ETH from the deposit pool when needed . Assessment: only a small share can be redeemed directly at this moment. Larger direct redemptions can become possible as pool liquidity returns, for example ETH returned by exiting validators , otherwise holders sell on a DEX, whose depth this analysis could not measure.
- Validator loss and slashing. Operator bonds absorb the first loss. A bond of 4 ETH per 32 ETH validator is thinner than the older 16 ETH bonds. The older FAQ example describes a loss beyond the bond as spread "across the entire network through a reduction in the relative value of rETH" . That FAQ describes the 8 and 16 ETH bonds, not the 4 ETH design, so how losses beyond 4 ETH are covered today is not confirmed. The only slashing condition the docs name is running the same validator keys on two nodes at once .
- New-code risk. Megapools and the Saturn 1 contracts were audited in January and February 2026 (Cantina 13 January, Hashlock and Bailsec January, Sigma Prime February). The project-hosted Cantina report records 2 Critical issues (marked fixed) and the Sigma Prime Saturn report records 1 High issue (marked fixed). The Bailsec report records 144 findings including 26 High, left fixes for several refactored contracts open pending re-review and does not claim high confidence that critical issues are gone. Assessment: smart-contract risk is higher than for the long-running 2021 to 2024 minipool code, until the fixes are confirmed and the code has more time live.
- Oracle committee trust. The oDAO reports validator balances and prices. A faulty or colluding majority could misreport the rETH rate. The documented guardrails cap this but are not independently verified here.
- Key concentration. The guardian is one key with undelayed control over the storage guardian setting . How much damage the guardian can do now that bootstrap mode is disabled is not established. The bootstrap-gated upgrade paths are likely narrower now, because they are tied to bootstrap mode.
setGuardianandsetDeployedStatuscarry only a guardian check and a zero delay, so they are not narrowed by that flag. - Liquidity and peg. The 24-hour rate update means DEX prices can move away from the contract rate between updates. The Maker feed looks stale (above).
- Complexity. Many contracts (minipools, megapools, queues, DAOs, rewards) share one storage registry. The audit-coverage mapping shows most named contracts have no attributed address in the source, so audit-to-deployment matching is limited.
- Tokenomics change. RPL operator rewards come from inflation that the docs plan to phase out, so operator incentives could shift.
Data Completeness
The core mechanics, TVL and on-chain control state are well covered; deployed-code audit matching, historical yield and DEX depth are not.
- Available: official Rocket Pool docs (protocol, liquid staking, node staking, glossary, FAQ, contracts), an Immunefi page, 18 audit documents with digests, DeFiLlama snapshot (2026-10-29), on-chain control reads for Ethereum at block 26294960, and extra on-chain reads at blocks 26296361 to 26296427 (megapool factory code, deposit pool and Security Council lookups).
- Missing or limited:
- No articles or Blockworks reports are in the sources.
- No dated historical TVL, rETH APR history or operator count.
- The on-chain read of the Ethereum contracts is partial: 51 official addresses were not read because the list was truncated; several DAO setting contracts have no resolved functions. Reads for Optimism, zkSync and Scroll are listed as not read or forbidden; the Arbitrum, Polygon and Base reads cover bridged token contracts only.
- Delay values for upgrade functions are unknown (null).
- The identity of the guardian key holder, and what the GMC and IMC committees control, are not stated in the sources.
- The Security Council row relies on a Japanese-language docs page for Houston. The current membership (1 member, a 2-threshold contract with 3 owners) was read on-chain at block 26296425, but who the three owners are is not stated in the sources.
- Audit provenance: the Cantina, Sigma Prime Saturn and Bailsec reports are project-hosted and their auditor provenance is unverified in the audit index, so their findings are reported as stated in the documents.
- Audit scope: the audit index could not confirm the code scope of the Sigma Prime Saturn review or the Bailsec report, and most contract names in audits have no attributed deployed address (70 such components). Audit coverage of the exact deployed megapool code is therefore not confirmed.
- The
feature-overview.mdpage uses pre-launch wording for Saturn 1; the project's timeline page lists the launch on mainnet as 18 February 2026 . - DEX liquidity depth for rETH is not measured (no candidates returned).
- Web research: two web questions (the live deposit pool address and the current Security Council members) found no address or roster on the pages returned. The deposit pool address and the council membership were instead resolved with read-only on-chain calls through
defres-web chain, and those RPC results are cited.
127 citations · click markers to view evidence
Code Security
Code Security Analysis: Rocket Pool
Rocket Pool has been reviewed by at least four independent firms on its Saturn upgrade (Hashlock, Bailsec, Cantina and Sigma Prime, January to February 2026), after a long history of earlier audits. The weak point is the link between audited and live code: no verified source for the live contracts is in the evidence, so deployed code was not compared with any audited version. Of 458 distinct audit findings, 235 are marked fixed by the auditors; 52 critical, high or medium findings are not marked fixed, and a repository check found the described code still present for several of them. The evidence supports a moderately confident reading: strong review depth, a modest bug bounty, and unproven alignment between audits and the live contracts.
Data Completeness
The ledger groups 561 reported findings from the included audits into 458 distinct issues; the findings ledger, the coverage map, the on-chain control file and the security page were read in full, and audit text was read for each quoted finding and for each audit's title page. Two duplicate audit copies were excluded from counting. Several gaps limit the conclusions:
- One ConsenSys Diligence report for the Redstone upgrade could not be read because the available copy is an image-only document.
- The audited revisions of the swap router repository and of a ConsenSys 2021 mirror repository are not published, so those audits cannot be tied to code.
- The coverage map lists no collected-source gaps, but its inventory is partial: 70 contract names from audits have no attributed live address, and only 2 addresses (both from documentation examples) were checked for source.
- 22 of the 52 not-fixed critical, high and medium groups could not be checked against code: the audited repository revision was not available for 9 (3 of these also hit the check budget), the check budget alone for 8, the described code was not found for 4, and the finding names no contract for 1. These groups carry only their audit status.
- Ten audit copies come from sources whose publisher is not verified as the auditor (project-hosted or discovered through a listing). Their content is used, with the caveat marked where a claim rests on them alone.
- Several Sigma Prime reports carry no date in the available copy, and the Cantina report lists issues only in part.
- Live on-chain state was read only for Ethereum, at block 26294960 (2026-10-29); other chains were not read.
Audit Coverage
The protocol has 16 included audit reports across five years, with the deepest recent coverage of the Saturn code. Total value locked is about $1.12 billion, all on Ethereum . The official protocol timeline dates the Saturn 1 mainnet launch to 18 February 2026 , and the first-party contract table lists megapool contracts on Ethereum mainnet . The feature overview still calls Saturn One upcoming, which is stale pre-launch wording . Saturn One is therefore the current major version, and the Saturn reviews below are reviews of it.
- ConsenSys Diligence, April 2021. Review of the original protocol and tokenomics update, with a mitigation update in May 2021; published on the auditor's own site. A second published copy exists (source reliability: unverified publisher), which this report treats as a duplicate.
- Sigma Prime, May 2021. Protocol review, version 2.0. The copy on the auditor's own site carries no date; the copy hosted by the project is of unverified provenance.
- Trail of Bits, September 2021. Security assessment of the protocol contracts.
- Sigma Prime, June 2022. Pre-merge upgrades. Provenance is unverified (project-hosted copy).
- ConsenSys Diligence, January 2023. Atlas release (version 1.2). Provenance is unverified (project-hosted copy).
- Sigma Prime, undated. Three further reports: the Atlas contract review, the swap router review and the ETH2 deposit mechanism review.
- ConsenSys Diligence, December 2023. Houston upgrade. Source reliability is unverified for this copy.
- ChainSafe, April 2024. Review of the Houston DAO contract update, 36 findings listed in the ledger.
- Sigma Prime, June 2024. Houston upgrade review. Provenance is unverified.
- Sigma Prime, September 2024. Houston hotfix diff review. Provenance is unverified.
- Hashlock, January 2026. Saturn contracts at commit
c9d9cbf; published on the auditor's own site. - Bailsec, January 2026. Saturn upgrade with three resolution rounds and a follow-up on the megapool design; 144 findings per the summary, 26 of them high.
- Cantina, January 2026. Managed review of the Saturn code. . The report is truncated, so only part of its finding list is visible.
- Sigma Prime, February 2026. Saturn upgrade review of a diff between two commits. Provenance is unverified.
Unaudited or weakly covered areas: the Redstone upgrade report is unreadable. Bailsec reports that a follow-up commit other audit providers had considered finalized still had new fund-theft and DoS issues . It states it cannot say with high confidence that no critical issues remain , recommends a third-party full audit and a bug bounty before deployment , and recommends a focused audit of the megapool network balance accounting .
Deployed code vs audits
The coverage map contains 2 components with an attributed address and 70 audit-named contracts without one, so deployed code was not compared with the audited version for any of them, and no share of identical, changed or unaudited lines can be given. The map reports no collection gaps, and its inventory is incomplete. The 2 addressed components are example addresses from documentation whose source is not verified. The first-party contract list names 8 mainnet contracts that the map does not include: rETH, rocketVault, rocketMegapoolManager, rocketMegapoolProxy, rocketNodeDeposit, rocketMinipoolQueue, rocketNetworkRevenues and beaconStateVerifier . Missing addresses do not show that these contracts are unaudited; they show that the link from audit to live code is not established.
| Component | Deployment | Status | Audit and revision | Criticality | Note |
|---|---|---|---|---|---|
| The node's fee distributor | Ethereum 0xa0bfbfc582f5814585f8455ed6d7b620ea9a9ee4 | Source not available (not verified) | None attributed | Peripheral | Address taken from documentation prose or an example; not a protocol core contract. Source not verified. |
| Example address in the German documentation | Ethereum 0xd3e317806456102d19db283b4c2201f0ef41a296 | Source not available (not verified) | None attributed | Peripheral | Address taken from documentation prose or an example; not a protocol core contract. Source not verified. |
Deposit | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Hashlock, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
MerkleDistributor | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RPL token contract | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Sigma Prime, May 2021; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketClaimDAO | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAONodeTrustedActions | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAONodeTrustedProposals | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAONodeTrustedUpgrade | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAONodeTrusted | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAOProtocolProposal | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDAOProtocolVerifier | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketDepositPool | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMegapoolDelegateBase | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMegapoolDelegate | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Sigma Prime, Feb 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMegapoolFactory | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example Bailsec, Jan 2026; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMerkleDistributorMainnet | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMinipoolBase | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMinipoolDelegate | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketMinipoolManager | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketNetworkBalances | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketNodeDistributorDelegate | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketNodeDistributor | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketNodeStaking | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketRewardPool | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Apr 2021 (second copy); revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketStorage | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Apr 2021; revision not compared | Critical | Deployed code was not compared with the audited version. |
RocketTokenRPL | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | Critical | Deployed code was not compared with the audited version. |
Oracle node voting | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Apr 2021 (second copy); revision not compared | High | Deployed code was not compared with the audited version. |
RocketAuctionManager | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | High | Deployed code was not compared with the audited version. |
RocketDAONodeTrustedSettingsMembers | No address attributed | Version not compared: no deployed address attributed | Named in an audit, for example ConsenSys Diligence, Jan 2023; revision not compared | High | Deployed code was not compared with the audited version. |
42 further rows are not shown: 32 high-criticality and 10 peripheral names, all without an attributed address. The map's inventory gaps repeat this for 47 contract names.
Independent code observations
- [OBS-1] (hypothesis, informational): the
RocketStorageregistry reports the guardian as address0x0ccf14983364a7735d369879603930afe10df21e, and the on-chain read at block 26294960 (2026-10-29) shows no contract code at that address, so it is a single key . The repository still sets the guardian to the deploying account in the constructor . The June 2022 Sigma Prime review (hosted in the docs) states the guardian was planned to move to a 2 of 3 multisig . Which powers the guardian retains after the deployment phase was not traced; that would decide whether this matters.
Findings by Severity
Of 458 distinct findings, 235 are marked fixed and are trusted without re-checking: 9 of 11 critical, 45 of 60 high and 37 of 72 medium groups. The groups below are the 52 critical, high or medium findings not marked fixed. A "Check" compares the finding with the latest public repository code and never with deployed code. Groups that rest only on audit copies with an unverified publisher are labeled as such.
Critical
- RocketDaoNodeTrusted - DAO takeover during deployment/bootstrapping (Critical; ConsenSys Diligence, Apr 2021) Reported in: ConsenSys Diligence, Apr 2021; ConsenSys Diligence, Apr 2021 (second published copy). — Audit status: not stated — Check: not checked: the audited repository revision was not available.
- RocketDAOProtocolSettings/RocketDAONodeTrustedSettings - anyone can set/overwrite settings until contract is declared deployed (Critical; ConsenSys Diligence, Apr 2021) Reported in: ConsenSys Diligence, Apr 2021; ConsenSys Diligence, Apr 2021 (second published copy). — Audit status: acknowledged — Check: not checked: the audited repository revision was not available.
High
- RocketRewardPool - Unpredictable staking rewards as stake can be added just before claiming (High; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "RocketRewardPool - Unpredictable staking rewards as stake can be added just before claiming"; ConsenSys Diligence, Apr 2021 (second published copy): "RocketRewardPool - Unpredictable staking rewards as stake can be added just before claiming and rewards may be paid to operators that do not provide a service". — Audit status: partially fixed — Check: not checked: the audited repository revision was not available.
- RocketStorage - Risk concentration by giving all registered contracts permissions to change any settings (High; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "RocketStorage - Risk concentration by giving all registered contracts permissions to change any settings"; ConsenSys Diligence, Apr 2021 (second published copy): "RocketStorage - Risk concentration: all registered contracts can change any settings (Major)". — Audit status: acknowledged — Check: not checked: the audited repository revision was not available.
- RPL slashing insufficient - node operators can withdraw stake to minimize slash exposure (High (estimated); ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "RPL slashing insufficient - node operators can withdraw stake to minimize slash exposure"; ConsenSys Diligence, Apr 2021 (second published copy): "RPL slashing can be circumvented via stake withdrawal after cooldown". The audit states no rating; this severity is an AI estimate, not the auditor's rating. Reason: header truncated; slashing cap can be bypassed via cooldown withdrawal, allowing node operators to avoid covering up to 32 ETH losses. (the report lists a closely related finding, node operators reducing slashing impact by withdrawing excess RPL, as fixed by this change; the estimated finding itself carries no stated status) — Audit status: not stated — Check: not checked: the audited repository revision was not available.
- Upgradeable contracts can block minipool withdrawals (High; Trail of Bits, Sep 2021) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- tx.origin in RocketStorage authentication may be an attack vector (High; Trail of Bits, Sep 2021) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) . Assessment: on Ethereum at block 26294960 (2026-10-29) the registry reports its deployment phase as finished , and the
tx.originbranch in the code applies only before that, so the exposure appears closed on mainnet; this does not change the code check above. - RocketMinipoolDelegate - Sandwiching of Minipool calls can have unintended side effects (High; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: not stated — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Multiple checks-effects violations (High; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: not stated — Check: not checked: the finding names no contract to locate.
- Upgraded Contracts Threaten Proposal Bond Liquidity (High; Sigma Prime Houston review, Jun 2024 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Increase of reduced.bond will break multiple transitions (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: partially fixed — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Lack of idempotency within deprecate will result in constant extension of expirationBlock (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Manipulation of ETH bonded by withdrawing bonded amount without finalizing the minipool (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: partially fixed — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- initialiseFeeDistributor function can be abused to inflate average fee (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Unclaimable rewards post-update due to mismatch of leaf (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: not stated — Check: a fix is visible in the latest repository code (rocket-pool/rocketpool@fef41a4); deployment not confirmed (latest repository) .
- ETH is locked in case of actionChallengeMake call by non trusted node (High; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) Members: ConsenSys Diligence Atlas review, Jan 2023: "RocketDAONodeTrustedActions - No way to access ETH provided by non-member votes"; Bailsec Saturn review, Jan 2026: "ETH is locked in case of actionChallengeMake call by non trusted node". — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Node Operators Can Claim RPL Stake Without Running A Node (High; Sigma Prime Atlas review (undated)) — Audit status: acknowledged — Check: a fix is visible in the latest repository code (rocket-pool/rocketpool@fef41a4); deployment not confirmed (latest repository) .
Medium
- Trusted/Oracle nodes can vote multiple times for different outcomes (Medium; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "Trusted/Oracle nodes can vote multiple times for different outcomes"; ConsenSys Diligence, Apr 2021 (second published copy): "Trusted/Oracle nodes can vote multiple times for different outcomes (Medium)". — Audit status: not stated — Check: not checked: check budget; the audited repository revision was not available.
- RocketDAO - personally identifiable member information (PII) stored on-chain (Medium; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "RocketDAO - personally identifiable member information (PII) stored on-chain"; ConsenSys Diligence, Apr 2021 (second published copy): "RocketDAO - PII stored on-chain (Medium)". — Audit status: acknowledged — Check: not checked: check budget; the audited repository revision was not available.
- Deployment - Docker containers running as root (Medium; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "Deployment - Docker containers running as root"; ConsenSys Diligence, Apr 2021 (second published copy): "Deployment - Docker containers running as root (Medium)". — Audit status: not stated — Check: not checked: check budget; the audited repository revision was not available.
- Deployment - Overloaded Ubuntu base image (Medium; ConsenSys Diligence, Apr 2021) Members: ConsenSys Diligence, Apr 2021: "Deployment - Overloaded Ubuntu base image"; ConsenSys Diligence, Apr 2021 (second published copy): "Deployment - Overloaded Ubuntu base image (Minor)". — Audit status: not stated — Check: not checked: the audited repository revision was not available.
- Minipool state machine design and pseudo-states (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- Sparse documentation and accounting complexity (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- RocketNodeDistributor - Missing extcodesize check in dynamic proxy (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- Kicked oDAO members' votes taken into account (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- RocketDAOProtocolSettingsRewards - settings key collision (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- RocketDAOProtocolSettingsRewards - missing setting delimiters (Medium; ConsenSys Diligence Atlas review, Jan 2023 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- Behavioural Inconsistencies In Protocol Settings Initialisation Code (Medium; Sigma Prime Houston review, Jun 2024 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- Lack of expiration state for upgrade allows for executing stale upgrades (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- overrideVote can be used to change voting direction (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Race condition for increase of proposalBond/challengeBond can result in unexpected lock/loss (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- No User-Side Slippage Protection On RocketDepositPool.deposit() (Medium; Sigma Prime Saturn review, Feb 2026 (source reliability: unverified publisher)) Members: Bailsec Saturn review, Jan 2026: "Lack of slippage during mint in case of fee change race condition"; Sigma Prime Saturn review, Feb 2026: "No User-Side Slippage Protection On RocketDepositPool.deposit()". — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Beaconchain slashing withdrawal delays allows debt accrual and repayment bypass (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- DoS of reward claim due to edge-case with two notifyExit calls (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Ambiguous behavior due to slashing after exit can result in malicious action and deliberate DoS of rewards (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Reward payout lag due to slashing will cause side-effects in follow-up distributions (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: partially fixed — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Call of distribute before any active validator will skew lastDistributionBlock and related share ratio calculation (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: partially fixed — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Loss of prestakeValue in case of missing stake call for 28 days (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Donation can force-slash Minipool owner (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Unfair reward distribution for period between [statusTime, promote] (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: partially fixed — Check: the code was rewritten; the check could not determine whether this finding still applies (latest repository) .
- Broken incentive model for fee recipient/coinbase address (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Multiple unexpected implications in case of contract update (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Edge-case in claimOutstandingETH may result in permanently locked funds (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: the described code was not found in the repository.
- Node with outstanding invitation cannot be kicked (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Node with outstanding invitation cannot be kicked (Security DAO) (Medium; Bailsec Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Missing guardrails for upgradeveto.quorum and upgrade.delay can affect protocol upgrades (Medium; Cantina Saturn review, Jan 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: not checked: check budget.
- Bond Requirement Used In Capital Dispersal Calculations Not Cached (Medium; Sigma Prime Saturn review, Feb 2026 (source reliability: unverified publisher)) — Audit status: acknowledged — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- Inaccurate RPL Inflation When Minting Multiple Intervals At Once (Medium; Sigma Prime protocol review (undated copy)) Reported in: Sigma Prime protocol review (undated copy); Sigma Prime protocol review, May 2021. — Audit status: open — Check: a fix is visible in the latest repository code (rocket-pool/rocketpool@fef41a4); deployment not confirmed (latest repository) . Conflict: the audit status is open, while the code check shows a fix; no later audit in the workspace confirms the fix.
- Unbonded minipools have proportionally smaller RPL requirement (Medium (estimated); Sigma Prime protocol review (undated copy)) The audit states no rating; this severity is an AI estimate, not the auditor's rating. Reason: No severity stated; estimated from reported collateral requirement bypass affecting economic security. — Audit status: not stated — Check: not checked: the described code was not found in the repository.
- RocketDAOProposal callable directly (Medium (estimated); Sigma Prime protocol review (undated copy)) The audit states no rating; this severity is an AI estimate, not the auditor's rating. Reason: No severity stated; estimated from reported ability to bypass the intended proposal flow. — Audit status: not stated — Check: the latest public repository code (rocket-pool/rocketpool@fef41a4) still contains the code this finding describes; deployed code was not compared (latest repository) .
- RPL token contract upgradeable with empty ABI (Medium (estimated); Sigma Prime protocol review (undated copy)) The audit states no rating; this severity is an AI estimate, not the auditor's rating. Reason: No severity stated; estimated from reported ability to upgrade a core token contract. — Audit status: not stated — Check: not checked: the described code was not found in the repository.
- Replaced DAO members can submit same exits to increase voting power (Medium (estimated); Sigma Prime protocol review (undated copy)) The audit states no rating; this severity is an AI estimate, not the auditor's rating. Reason: No severity stated; estimated from reported ability to inflate voting power through duplicate exits. — Audit status: not stated — Check: not checked: the described code was not found in the repository.
Low / Informational
The ledger holds 141 low-severity groups (73 fixed, 68 not fixed) and 174 informational groups (71 fixed, 103 not fixed). Low and informational findings are not individually checked against code. The Hashlock report lists one medium finding, resolved . The Cantina digest reports two critical issues in its visible list, marked fixed .
Recurring Patterns
Several issues appear in more than one audit, which points to design choices rather than one-off mistakes.
- Duplicate publication of the 2021 ConsenSys review. Nine of the not-fixed groups appear in both copies of the same report, so they count once; the second copy has an unverified publisher.
- Same Sigma Prime review published twice. The May 2021 review appears as a dated project-hosted copy and an undated auditor-hosted copy; the RPL inflation finding is in both.
- Ether sent to the challenge function is locked. ConsenSys (January 2023) and Bailsec (January 2026) both report that ETH sent to
actionChallengeMakecannot be recovered. In the repository check both members still show the described code. - Deposit slippage. Bailsec and Sigma Prime (February 2026) both raise slippage in the deposit path; both are acknowledged and the check shows the code present.
- Concentrated storage and upgrade powers. ConsenSys (2021) and Trail of Bits (2021) report that any registered contract can change any setting and that governance can replace contracts that gate minipool withdrawals. Both are acknowledged. The repository check still finds the code behind the Trail of Bits upgrade-power finding; the ConsenSys storage finding could not be checked.
- Bootstrapping and deployment-phase powers. The 2021 critical findings on DAO takeover and open settings during deployment are not marked fixed; the audited repository revision was not available to check.
Code Quality Assessment
The audit record is deep and recent, with a fix rate that is high for critical issues and lower for medium ones. Findings are fixed for 235 of 458 groups overall: 9 of 11 critical, 45 of 60 high and 37 of 72 medium. Both unfixed critical findings date from 2021 and concern the bootstrapping phase. The registry reports its deployment phase as finished , which closes the settings window only. Whether the DAO-takeover finding still matters depends on the powers the guardian retains, which were not traced. The Saturn reports are the main concern: Bailsec found 26 high findings, and the repository check still finds the code for several acknowledged or partly fixed Saturn items, including the challenge deposit, bond handling and megapool distribution logic. Assessment: the code base is audit-intensive but changes quickly, so fixes landed after the January 2026 reports are the key unknown.
Assurance after audit is moderate. The Immunefi bug bounty pays up to $150,000 for critical bugs and has run since 8 September 2021 . Source verification for the live contracts is not shown. A third-party Exponential snapshot is supporting evidence only; it notes review by several experienced auditors .
Score: 4.3/10
Rubric security-score-v2.5; data completeness 1.00
- Audit coverage
- 5.50 points
- Audit depth
- 7.50 points
- Unresolved findings
- 0.00 points
- Assurance
- 5.50 points
- Critical · not stated
- 1 findings
- Critical · acknowledged
- 1 findings
- High · not stated
- 4 findings
- High · partially fixed
- 3 findings
- High · acknowledged
- 8 findings
- Medium · open
- 1 findings
- Medium · not stated
- 7 findings
- Medium · partially fixed
- 3 findings
- Medium · acknowledged
- 24 findings
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Audit coverage (computed) | L2 | supported | 0.35 | 5.50 |
| Audit depth | L3 | supported | 0.20 | 7.50 |
| Unresolved findings (computed) | L0 | supported | 0.30 | 0.00 |
| Assurance | L2 | supported | 0.15 | 5.50 |
123 citations · click markers to view evidence
Economic Security
Economic Security Analysis: Rocket Pool
Rocket Pool is an Ethereum liquid staking protocol with about USD 1.12B TVL, all on Ethereum mainnet. Its main economic strengths are that the returns on rETH come from staking rewards and that node operators' own ETH absorbs the first losses. Its main weaknesses are that the rETH exchange rate is pushed by a ten-member oracle committee, direct rETH redemption liquidity is under 1% of supply, and no insurance or reserve fund was found. The evidence is solid on mechanics and on-chain state. It is thin on holder concentration, DEX depth and Saturn 1 loss-handling details.
TVL & Liquidity Profile
The protocol holds roughly USD 1.12B on a single chain, so chain concentration is total and L2 exit risk does not apply to the TVL.
- DeFiLlama snapshot, collected 2026-10-29: TVL is 1,118,047,553 USD, all on Ethereum . The category is Liquid Staking. The 7-day change in the same snapshot is -11.7% . Pair this with the snapshot date; it is one capture, not a trend.
- DeFiLlama's separate staking series is 11.44M USD . This is about 1% of TVL.
- The DefiLlama token composition is 100% WETH . The WETH node is unresolved in the dependency graph .
- Chain profile: Ethereum only, with no L2BEAT stage or exit window because it is the base chain . The integrations table lists rETH on Base and on other chains, but those tokens carry no counted TVL. The on-chain reader did not read two of those chains and was refused on Optimism .
- rETH on-chain state, Ethereum block 26294960: supply is 308,566 rETH at an exchange rate of 1.1746 ETH per rETH. That is about 362,000 ETH (supply times rate, calculated here) .
- Direct redemption liquidity: the rETH contract and deposit pool hold about 2,646 ETH of collateral, a collateral rate of 0.73% . The deposit pool alone holds 16.3 ETH.
- DEX depth: not measured. The depth probe found no candidate assets . The 2% and 0.5% depth for rETH are therefore gaps, not zero. Docs name a Balancer composable stable pool and a Uniswap ETH-rETH pool as secondary venues . Pool pages captured on 29 October 2026 (source reliability: unverified publisher) show total pool value, not depth at 2% price impact: a Uniswap v3 0.05% pool at USD 44.1K , a Uniswap 0.01% pool at USD 748.9K and a Balancer v3 pool at USD 4,341,143 . Snapshot dates on those pages were not confirmed, and a Rocket Pool forum post (source reliability: community publication) says mainnet rETH liquidity fell from over USD 30M to about USD 2.5M after the Balancer hack . These figures are unconfirmed leads from weak sources; market-wide depth at 2% price impact stays unresolved.
- Holder concentration: an Etherscan holder page captured on 29 October 2026 (source reliability: unverified publisher) gives the top 10 holders as 29.66% of supply . The largest holder is an Aave rETH contract at 11.3097% . Five of the top ten (7.73% combined) are unlabelled addresses. The figure comes from one weak publisher and combines lending markets, vaults and a gateway with possible individual holders, so it is a lead, not a verified measure.
- Native exit queue: the docs state no maximum delay for redeeming rETH with the protocol. They say the exit depends on pool liquidity (see Solvency).
Oracle Risk
The rETH/ETH rate and the RPL price are pushed by the Oracle DAO (oDAO) rather than computed on-chain from verifiable balances. A rate-change limit is specified in a governance proposal and discussed in an audit, but the live limit was not read, so trust in the ten-member committee is the safeguard the sources establish.
- Source of the rate: the docs say Oracle Nodes are "A special node run by trusted members of the Rocket Pool community that relays information from the beacon chain (such as total staking rewards and validator exit status) back to the execution layer so the Rocket Pool smart contracts can function correctly." . The rETH/ETH rate updates about every 24 hours .
- Committee size on-chain (Ethereum block 26294960): the oDAO contract reports 10 members, a minimum of 3, bootstrap mode disabled and a vote quorum of 5.1 votes . In practice 6 of 10 members must agree. An official docs page says submissions need a 51% quorum . That page's sample CLI output shows 8 members; it is an illustration, and the on-chain read of 10 is the current figure.
- Specified bounds on the rate (RPIP-61, a governance proposal, not a confirmed live setting): the initial maximum rETH change per update is 2% . The initial rule also blocks a balance update if less than 95% of the submission interval has passed . The RPIP page is the specification and 2% is its initial value; the current on-chain delta was not read, so the live limit is not confirmed.
- Audit evidence on the guardrail: a Sigma Prime Saturn 1 review (publication host not verified as auditor-controlled, so lower confidence) found that the ETH delta was limited but rETH supply was not . The report marks it resolved in commit 399c7ba .
- RPL price: the oDAO also submits the RPL price. The on-chain read shows a price at block 26289290 against a latest reportable block of 26265600 . The Cantina review found that the RPL price update frequency is not enforced on the oDAO . Cantina's report is also of unverified provenance.
- Downstream feeds: Chainlink and Maker publish rETH feeds for integrators . These are used by other protocols; the staking contracts do not depend on them.
- Fallback if the oracle fails: no fallback is documented in the workspace. Assessment: a stalled oDAO would leave the rate stale rather than wrong, while a compromised quorum could push a wrong rate. If the RPIP-61 limit (initially 2% per update) is live, a single update is bounded; the live value is unread, so the size of a possible wrong rate is not established.
- Historical oracle losses: none recorded in the workspace. The only DeFiLlama hack entry is not an oracle event (see Incentive Alignment).
Flash Loan & MEV Exposure
The main MEV vector is trading around the daily rate update. A 2021 audit rated it critical and the audit's resolution text describes a deposit delay as the fix, but the deployed delay was not read.
- Rate-update sandwich: an April 2021 ConsenSys Diligence audit rated "6.2 RocketTokenRETH - sandwiching opportunity on price updates Critical ✓ Fixed" . The fix introduced a delay of about one day between an rETH deposit and a later transfer or burn . The resolution text says the delay was in a pending pull request at that time and later says a zero delay is not possible . Assessment: if that delay is live, a same-block flash loan cannot complete a deposit-then-burn round trip; the workspace does not show the deployed setting, so this is not confirmed.
- Bailsec (Saturn 1, unverified provenance) re-raised the point as "Reliance on executeBalance can result in MEV" . That separate Saturn 1 issue was rated informational and acknowledged, citing the fee and protocol size.
- Static liquidity buffer: Sigma Prime found that under heavy deposits the liquidity buffer can be breached between price updates . The project acknowledged it. This is a liquidity risk, not a flash-loan exploit.
- Validator-level MEV: the Cantina review lists "Compromised oDAO can apply MEV theft penalty to arbitrary validators" . It was acknowledged. This is a trusted-committee risk, not a permissionless MEV attack.
- Liquidations: rETH holders have no borrower position, so liquidation front-running does not apply to the rETH product.
Tokenomics Risk: Yield, Leverage & Reflexivity
rETH holders earn staking rewards, with no own-token emission in their return. Node operators run 8x-levered validators and earn partly in RPL, and RPL is a secondary collateral. The RPL share of backing is not measured.
- rETH return source: Beacon Chain rewards, priority fees and MEV rewards raise the rETH/ETH rate . The cumulative amount is visible on-chain: the rate was 1.1746 at block 26294960, against a documented starting point of 1 ETH = 1 rETH . No APY figure is relied on.
- Revenue split under Saturn 1 (live since 18 Feb 2026 per the official site ): node operators receive a 5% base commission share and RPL stakers a 9% voter share. The remaining ETH revenue goes to rETH holders . The pDAO share is 0% and adjustable. The pDAO governance can adjust each share, so these splits can change.
- Own-token emissions: RPL inflation is paid mainly to node operators, not to rETH holders. The docs say that the majority of RPL inflation goes to node operators . The on-chain RPL contract shows 22.96M supply and a daily interval rate of 1.00013368, about 5% a year (calculated here) . The share of RPL emission within node operators' total paid yield is not cited, so the split is a gap.
- Leverage in the node operator position: a megapool validator uses 4 ETH of the operator's capital and 28 ETH from liquid stakers . That is 8x leverage on the operator's capital. The unwind rule is stated: if the validator ends up with less than 32 ETH, the docs say "all of the lost ETH is coming out of your share" . The docs do not state here what happens to a loss larger than the 4 ETH bond. rETH holders hold no leveraged position.
- Own token as collateral: RPL "can serve as secondary collateral (e.g., against slashing)" . It is no longer required for megapool validators . The RPL share of total backing or collateral is not measured: DeFiLlama's 11.4M USD staking series is a separate line and does not isolate RPL collateral. User claims on rETH are described as ETH deposits and ETH redemptions, not minted or redeemed against RPL.
- Reflexive loss path: an RPL price collapse would weaken only the secondary collateral and node-operator incentives. rETH would still be backed by staked ETH. The workspace does not show an RPL-driven loss event.
- Holder and voter concentration are outside this section.
Solvency, Liquidation & Bad Debt
rETH is backed by ETH staked in validators, with a rate reported by the oDAO. Direct redemption is limited to a small buffer, and no insurance fund was found in the sources searched.
- Backing: the docs define rETH value as total staked ETH plus contract balance plus the rETH share of rewards, divided by rETH supply . The staked balance is reported by the oDAO from the Beacon Chain, so it is team-published rather than proven on-chain.
- Exit path: redemption with the protocol works only while ETH is in the pool. The docs say: "It's possible that if node operators have put all of the staking pool to work on the Beacon chain, then the liquidity pool won't have enough balance to cover your unstaking." . Measured redemption collateral is 0.73% of supply (see TVL section). Larger exits depend on DEX liquidity (not measured) or validator exits. No maximum delay is stated.
- Docs buffer target: the Saturn 1 docs describe a buffer of 1% of rETH TVL in the rETH contract before surplus flows to the deposit pool . The Sigma Prime review refers to a 10% buffer, so the buffer wording differs between sources. The measured 0.73% is below the docs' 1% threshold.
- Loss handling before Saturn 1: the operator's bond and RPL pay first. Any remainder "is spread across the entire network through a reduction in the relative value of rETH" . That wording describes the older 8 or 16 ETH minipool model. For Saturn 1 megapools, the docs say a loss comes out of the operator's share , but a megapool loss waterfall beyond the 4 ETH bond was not found in a current docs page, so who absorbs a larger loss under Saturn 1 is not confirmed.
- Exit-accounting failure path: the project's security page discloses that if a node operator fails to notify an exit and the Oracle DAO also fails to challenge the megapool before the funds arrive, a call to
distributetreats the validator's principal as rewards and distributes it incorrectly . The team classes this as intended behavior and relies on the challenge process . Assessment: a high-impact operational path that depends on the trusted committee acting; no realized shortfall is recorded. - Legacy RPL slashing path: a Hashlock Saturn 1 finding showed that minimum legacy-stake checks could make an RPL slash revert, so no penalty applied if a node held only the minimum stake . The report marks it resolved. Slashed RPL is sent to the auction manager . This is a slashing mechanism, not a funded reserve of known size.
- Liquidation and bad debt: not applicable to rETH holders, since there are no borrower positions. No recorded bad-debt events were found. Absence of a record is not proof of no loss.
- Loss cover: no insurance, reserve or compensation fund was found. The docs use the word insurance for the operator's RPL bond , which is collateral, not a separate fund.
- Exit halts: no exit halt or pause is reported in the workspace at the research date.
Incentive Alignment
Operators have 4 ETH at risk per validator and share revenue with RPL stakers, so incentives point toward good operation. The weak points are thin bond relative to validator size and the lack of a loss-cover fund.
- Operator skin in the game: the 4 ETH bond plus RPL stake (secondary) is small against 32 ETH. Operators earn a 5% commission on the 28 protocol ETH . The docs state the max RPL yield applies up to 15% of borrowed ETH .
- Mercenary capital: No points or emission campaign tied to rETH was found in the sources searched, so the share of TVL linked to one is not established. RPL staker rewards depend on governance-adjustable splits and on RPL inflation planned to phase out with Saturn 2.
- Death-spiral risk: rETH is not described as minted or redeemed against RPL, but RPL serves as secondary collateral and its share is not measured, so an RPL-related loss path is not ruled out. Exponential's curated snapshot lists "No death spiral concerns" .
- Loss cover funds: none found. Web research found no captured page describing one. Pause and cooldown status was not assessed because no fund was found.
- Past incidents: DeFiLlama lists one Rocket Pool record, 2023-03-30, USD 3.8M, "Phishing" . A secondary report (source reliability: unverified publisher) attributes it to a phishing attack on one holder's wallet, around 85,900 RPL sold for ETH . That attribution is unconfirmed by a stronger source, and DeFiLlama records the event against the protocol, so the scope is unsettled. No protocol payout or compensation statement was found, so no payout row is recorded.
- Bug bounty: the Immunefi program pays up to USD 150,000 for critical bugs, 10% of funds affected . That is about 0.013% of TVL, a small incentive against a USD 1.1B pool.
Economic Attack Vectors
- Oracle committee collusion: 6 of 10 oDAO members could push a rate (any limit would be the RPIP-61 specification, initially 2% per update; live value unread), and the audits flag oDAO-compromise scenarios (rETH supply limit, MEV penalty power). The impact would be a mispriced rETH exchange rate; the workspace has no record of such an event, so the likelihood is not quantified. Even if the initial 2% delta is live, it would bound a single update, not repeated updates. A future check of the on-chain delta value would narrow this.
- Redemption squeeze: under stress, direct redemption covers about 2,646 ETH (0.73% of supply). Most holders would exit via DEX, where depth is unmeasured. A discount of rETH to its protocol rate is a plausible stress outcome. This is an assessment based on the measured collateral rate, not a recorded event.
- Exit notification failure: if an operator does not notify an exit and the Oracle DAO does not challenge, validator principal can be paid out as rewards (see Solvency). The stated protection is the Oracle DAO challenge, which also fines the operator.
- Missed-slot withdrawal proof: the security page describes a case where a missed Beacon slot lets a partial withdrawal be relabeled so megapool capital is finalized early. The team calls it a non-issue because the operator has a 6,750-epoch window to submit the real proof . This is the team's own likelihood judgment, not a measured probability.
- Slashing or operator loss above the bond: under the older minipool rule, a remainder is spread across rETH holders. Assessment: a similar spillover is plausible for megapool losses above the 4 ETH bond, but the current rule was not found. No cited cap or fund covers this.
- Price manipulation of RPL: affects only secondary collateral and legacy RPL mechanics; the Cantina note on RPL price frequency shows the oDAO can update it at any time.
- Bridge-related: the docs list rate providers for rETH on Optimism, Arbitrum, Polygon and zkSync Era . These tokens carry no counted TVL, and where those rates come from and how they are configured was not reviewed.
- Composability: Aave's rETH contract holds the largest share of supply (11.3%, see TVL), so an rETH rate error would reach that lending market first. rETH is used as collateral elsewhere ("you can ... use it as collateral" ). A stale or manipulated rate would propagate to those protocols. Their exposure is not quantified here.
- Dependencies: the oDAO oracle set, the node operator set and validator performance, and pDAO/Security Council upgrade control. The dependency graph lists no scored protocol dependencies and no warnings . Ethereum base-layer staking is the native dependency.
Aggregator Scores
- CertiK Skynet (captured 2026-10-24, curated snapshot): score 77.42, grade BBB, rating "average" . It is a general security score and does not test redemption liquidity or oDAO trust.
- Exponential (archive capture 2026-03-11, curated snapshot): risk score 9.0, rating "Best" . It also states a timelock of at least 48 hours and a multisig of at least 4 signers; governance controls are assessed elsewhere and were not re-verified here. It is older than seven months at the research date. Its text says Oracle DAO consensus needs 9 of 16 members , which does not match the on-chain read of 10 members and a vote quorum of 5.1 votes, so that description looks outdated. It also says there are no documented hacks, while DeFiLlama lists the 2023 phishing record; the two cover different scopes.
- DeFiSafety: no file in the workspace.
- DeFiLlama: TVL snapshot of 2026-10-29 above.
- Comparison: both curated ratings are favorable and agree with the mechanism review on audit depth; the RPL secondary-collateral share behind their death-spiral view is unmeasured. They do not capture the 0.73% direct redemption buffer or the oDAO rate trust, which this analysis treats as the main economic risks.
Data Completeness
Available: DeFiLlama TVL and composition, on-chain reads of rETH, RPL, oDAO and price contracts at Ethereum block 26294960, official docs on mechanics, audits including three Saturn 1 reports, hacks record, and several web research rounds, including the project's security page and holder and pool pages.
Missing or limited:
- rETH 2% price-impact depth (the depth probe returned no candidates and the captured pool pages give total pool value only, with undated snapshots) and an independently verified holder breakdown (the one top-10 figure comes from a single weak publisher).
- The current on-chain value of the RPIP-61 maximum rETH delta and the deposit-delay setting.
- Current Saturn 1 penalty and loss-waterfall rules (docs captured describe the older minipool model for the residual-loss rule).
- Revenue amounts in USD for rETH and for node operators, and the RPL emission share of operator yield.
- Compensation for the 2023 phishing incident: no statement found.
- Provenance of the Saturn 1 audit copies (Sigma Prime, Bailsec, Cantina) is not verified as auditor-controlled; findings cited from them carry lower confidence. The project's own security page lists Saturn 1 audits by Cantina, Bailsec, Sigma Prime and Hashlock , which supports that the reviews exist but does not verify the copies.
- Link between audited commits and the deployed Saturn 1 contracts was not checked, so fixed findings are historical mitigation evidence.
- Chain reads for two L2 chains were not performed and Optimism was refused.
Confidence: moderate (about 0.6). Mechanics and on-chain state are well evidenced. Concentration, depth and loss-waterfall details are not.
Score: 5.3/10
Rubric security-score-v2.5; data completeness 0.90
- Solvency redemption
- 5.50 points
- Oracle design
- 5.50 points
- Dependency risk
- 5.50 points
- Liquidity concentration
- 3.50 points
- Yield design
- 5.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Solvency redemption | L2 | supported | 0.35 | 5.50 |
| Oracle design | L2 | supported | 0.25 | 5.50 |
| Dependency risk | L2 | supported | 0.15 | 5.50 |
| Liquidity concentration | — | not disclosed | 0.10 | 3.50 |
| Yield design | L2 | supported | 0.15 | 5.50 |
74 citations · click markers to view evidence
Team & Transparency
Social & Team Transparency: Rocket Pool
Rocket Pool has a publicly named founder and general manager, a named operating company, and a long record of published audits and governance proposals. The only incident the hacks dataset attributes to it is a 2023 phishing loss of about USD 3.8M. Weak sources attribute that loss to an individual RPL holder's wallet, not the protocol contracts, but no first-party statement confirms this. Evidence on team identity and disclosure is strong. Evidence on commit-level GitHub activity, social audience size and the company's legal jurisdiction is missing.
Rocket Pool is an Ethereum liquid staking protocol (rETH is its liquid staking token). DefiLlama lists it on Ethereum only, with TVL of about USD 1.12bn .
Team Identity
The team is publicly identified: a named founder, a named general manager, other named contributors and a named company, but the company's jurisdiction is not stated in the available sources.
- David Rugendyke (founder). The protocol FAQ says Rocket Pool "was originally started in late 2016 by David Rugendyke, a senior developer with over 18 years commercial experience and a computer science background" . The official protocol history dates his start of work to December 2016 .
- Darren Langley (General Manager). His delegate profile states he is "the General Manager of the Rocket Pool core team" . The official history lists podcast and conference appearances by him, including Bankless with Rugendyke in December 2021 .
- Other named contributors. The RPIP index (the public proposal list) names Kane Wallmann as co-author of several final core proposals, such as "Implementation of an On-Chain pDAO" . Other names appear only as podcast guests or proposal authors; their roles are not described in the sources.
- Legal entity. The site footer names "Rocket Pool Pty Ltd" . A ChainSafe audit says this company commissioned the April 2024 Houston review . The sources do not state the company's registration jurisdiction. A web search for it also returned nothing. The "Pty Ltd" form suggests an Australian company, but that is an inference, not a sourced fact.
- DAO structure. Governance runs through the protocol DAO (pDAO, voted by node operators and RPL stakers), the Oracle DAO and a Security Council, described under Transparency Indicators.
- Operating history. The team is not anonymous. The official history shows the project operating since 2016, with a first investment from Consensys Ventures in May 2018 .
Track Record
Rocket Pool has run on Ethereum mainnet since 9 November 2021 and has delivered several major upgrades; the one recorded incident is a 2023 phishing loss of RPL tokens whose victim is not confirmed by any first-party source.
- Operating history. The official history states "Officially launched on Mainnet!" on 9 November 2021 . Code derives the family and version ages from the DefiLlama listing date.
- Delivery. The history dates Redstone (29 August 2022), Atlas (18 April 2023), Houston (17 June 2024), Saturn 0 (28 October 2024) and Saturn 1 (18 February 2026) as mainnet launches. One example: the Saturn 1 and Saturn 2 tokenomics proposal passed governance on 22 August 2024 and Saturn 1 went live on 18 February 2026 . That is about 18 months from vote to launch. The sources do not include a roadmap with promised dates, so adherence to dates is not measured.
- Core contracts (current version). The official contracts page lists the deposit contract and the vault. . The same page lists older deposit pool versions (v1.0, v1.1) separately.
- Incident in the hacks dataset. DefiLlama records a 30 March 2023 "Phishing" incident for Rocket Pool with an amount of USD 3,800,000 . A U.Today article (source reliability: unverified publisher) says about 85,900 RPL were withdrawn from a wallet that "most likely belongs to the founder of the POAP protocol" . A Lookonchain post on X (source reliability: unverified publisher) reports that 85,898 RPL worth $3.83M were transferred out by a phishing-labelled address . A Binance Square post (source reliability: unverified publisher) says the RPL was taken from a named individual, not from the protocol . If these accounts are right, a single holder's wallet was phished and the Rocket Pool contracts were not breached. These are secondary attributions only. U.Today states that the wallet owner had not commented . A web check of rocketpool.net, dao.rocketpool.net and the Rocket Pool X account found no Rocket Pool post-mortem or statement on the incident, so the victim and the protocol's role stay unconfirmed. The incident is recorded as a major loss because it exceeds USD 1M and no compensation is documented.
- Other incidents. No protocol-contract exploit appears in the hacks dataset for Rocket Pool or its family. The dataset is a snapshot, so this is not proof of absence.
- Stress events. A community forum post (source reliability: community publication) says that after the Balancer hack, mainnet rETH liquidity fell from "$30M+" to about "$2.5M+" . The post describes a liquidity loss linked to another protocol, not a Rocket Pool exploit; no independent source confirms the drop or its cause, and no rETH depeg was measured.
- Emergency pauses. The governance page says the Security Council can pause only, citing recent examples of disabling solo migration, node deposits and node registrations . The dates and duration of those pauses are not given, so the sources do not establish whether any pause or outage exceeded 24 hours, and no complete outage or exit-halt history is in the sources.
- Depeg history. The code-owned token check found no measurable candidate token, so no depeg history for rETH (
0xae78736Cd615f374D3085123A210448E74Fc6393on Ethereum, named in the contracts page) is available . Theassetslist is therefore empty.
Social Presence
Social data in the sources is thin: official Medium updates are regular, but commit counts and audience sizes are not available.
- Medium. The Rocket Pool publication shows bi-weekly updates, with a latest post dated 6 October 2026 and "Last published Oct 12, 2026" (source reliability: community publication) . The same page lists a 22 September 2026 update, which is consistent with a two-week cadence.
- GitHub. The official site links a public GitHub organisation . The sources contain no commit dates or contributor counts, so commit frequency over the last 12 months is not measured. A web check for commit data returned none.
- DAO forum. One thread in the sources, on rETH liquidity recovery, ran from 17 December 2025 to 5 January 2026 with replies from a Balancer representative and a vault operator (source reliability: community publication) . The Balancer representative's offer of incentives predates a later Balancer DAO decision: a captured Balancer pool page says Balancer DAO approved an orderly winddown, with most pools moving to withdrawals only on 30 October 2026 (source reliability: unverified publisher) . This may reduce one source of rETH liquidity. The sources give no forum activity measure beyond this thread. The official history lists periodic roadmap update posts from January 2025 to October 2025 .
- Twitter/X, Discord, Telegram. Links exist on the official site, but follower counts, member counts and discussion quality are not in the sources. Insufficient data for this assessment.
Transparency Indicators
Documentation, contract addresses and audit reports are published first-hand, and governance is on-chain, but no treasury or revenue reports are in the sources.
- Address list. The official docs publish key mainnet contracts and the full list of active deployed contracts . The same page also lists some older entries next to current ones, so readers should confirm the active version through the
RocketStoragecontract. - Audits. The official security page lists reports by Sigma Prime, Consensys Diligence, Trail of Bits, ChainSafe, Cantina, Bailsec and Hashlock, covering releases from the 2021 mainnet launch to Saturn 1 in 2026 . Audit depth and coverage are assessed in the code-security report, not here.
- Known issues. The security page documents accepted issues "for transparency" .
- Bug bounty. An Immunefi programme is listed, live since 8 September 2021 with a maximum bounty of USD 150,000 and a last update of 2 September 2026 . The Rocket Pool security page lists "No KYC" while the Immunefi page lists "KYC required", a small inconsistency .
- Governance transparency. Every change starts as a public proposal (RPIP). The pDAO vote uses vote power equal to the square root of staked RPL, capped at 150% of bonded ETH . Contract upgrades have a one-week delay and a Security Council veto with a 33% quorum, which must publish an explanation within 24 hours .
- Financials. The docs name grant and incentive committee multisigs (GMC and IMC) . The sources contain no treasury, revenue or expense reports.
- Open source. The official site links a public GitHub organisation. Verified-source status of the deployed contracts is assessed in the code-security report, not here.
Blockworks (ex-Messari) Article Freshness Rules
The workspace has no Blockworks (ex-Messari) reports or other articles, so no freshness rule applies.
Red Flags
- No team anonymity concern: the founder, general manager and company are named.
- The 2023 phishing loss of USD 3.8M is the only recorded incident. Weak sources point to a single holder's wallet, but no first-party post-mortem or statement exists, so the cause and victim are unconfirmed.
- The legal entity's jurisdiction is not stated in the available sources.
- The Immunefi page and the Rocket Pool security page disagree on whether the bounty requires KYC.
- The maximum bounty of USD 150,000 is small relative to TVL of about USD 1.12bn.
- No treasury or revenue reporting is in the sources.
Positive Signals
- Named founder with documented professional background and a project history dating to 2016.
- Mainnet live since 9 November 2021 with repeated upgrades delivered through public proposals.
- Audit reports published for the mainnet launch, Redstone, Atlas, Houston and Saturn 1 releases. The audit list shows no report for Saturn 0 (28 October 2024), so the record is not complete for every release.
- On-chain protocol governance with a published process, a delay on upgrades and a veto limited to stated cases.
- Regular bi-weekly Medium updates through October 2026.
Data Completeness
Available: official site pages, documentation, the proposal index, the Immunefi page, DefiLlama hacks and TVL data, one forum thread and three weak publications on the 2023 incident (a news article, an X post and a Binance Square post).
Missing: commit dates and contributor counts, Twitter/X, Discord and Telegram metrics, treasury reports, the company's jurisdiction, rETH depeg measurements, a complete outage and pause history, a first-party statement on the 2023 incident and any Blockworks reports. This analysis uses collected pages, not live social media scraping.
Score: 5.7/10
Rubric security-score-v2.5; data completeness 1.00
- Team identity
- 7.50 points
- Disclosure
- 3.50 points
- Development activity
- 5.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Team identity | L3 | supported | 0.45 | 7.50 |
| Disclosure | L1 | supported | 0.35 | 3.50 |
| Development activity | L2 | supported | 0.20 | 5.50 |
Track record: 6.7/10
Rubric security-score-v2.5; data completeness 0.80
- Operating history
- 9.50 points
- Incident record
- 6.25 points
- Resilience
- 3.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Operating history (computed) | L4 | supported | 0.30 | 9.50 |
| Incident record (computed) | L2 | supported | 0.50 | 6.25 |
| Resilience | — | not disclosed | 0.20 | 3.50 |
37 citations · click markers to view evidence
Governance
Governance Analysis: Rocket Pool
Rocket Pool is governed by two on-chain DAOs rather than a single admin multisig. Protocol upgrades are proposed by the Oracle DAO (oDAO, a permissioned node set) and can be vetoed by a Security Council whose only member was a team multisig in January 2026. The official process page lists a one-week upgrade delay after the oDAO vote, and the guardian's bootstrap powers (including bootstrap upgrades) are documented as disabled. A user should still weigh two gaps. The guardian address that RocketStorage reports is a single key with zero-delay control over the guardian and deployed-status settings, and its other remaining powers could not be resolved. Voter concentration on the Protocol DAO (pDAO) could not be measured. Confidence is moderate: the delay and veto design is well documented, but current Security Council signers and vote concentration are not confirmed.
Governance Structure
Governance is a hybrid of two on-chain DAOs plus a team-run emergency layer.
- Protocol DAO (pDAO): it is made up of RPL holders, and staked RPL gives voting power . Voting power is
VP = sqrt(min(rplStake, bondedETH × maxStakePct / rplPrice)). One document adds "(min 28 days)" to the staked RPL . The Saturn audit appendix computes voting power from snapshots , and the node documentation uses 28 days as the wait after initiating unstaking . A minimum stake age for voting is therefore not established. The square root dampens large holders. - Proposals: any node with non-zero voting power can submit or vote on a pDAO proposal at any time . Proposals change pDAO settings, spend treasury funds or change Security Council membership. Execution uses an optimistic fraud-proof scheme, with a 7-day challenge period . The official governance process page gives a 15% quorum for on-chain pDAO proposals and says individual node operators can override their delegate's vote . The delegates page lists node operators to whom voting rights can be delegated .
- Oracle DAO (oDAO): it is made up of the oracle node operators . The oDAO proposes contract upgrades. Its own votes use a 51% quorum setting .
- Security Council (SC): pDAO elects and removes its members, and it can act without delay . The SC has veto power over upgrades .
- Timelock: an upgrade passes a vote delay and then an
upgrade.delay. In the January 2026 review the team states: "So the minimum time between a NO seeing an upgrade proposal and it being executed is currently 2 weeks." - Emergency bypass: the SC can change a short list of flagged parameters without delay. Cantina found that
network.submit.prices.enabledhad been implemented as SC-changeable without delay although the specification did not flag it; the team records "Fixed in commit 6c31f26e. Added missing permission." The flagged-parameter list is not reproduced in the sources read. - Vote concentration: not measured. The measured file lists the Snapshot space
snapshot:rocketpool-dao.ethwithstatus: unsupported_typeand null shares . No protocol document read names that Snapshot space as the deciding venue, so no voting venue is claimed.
Admin Key Analysis
Documents say bootstrap upgrade powers are disabled, but a single-key guardian address and a team-run Security Council remain.
- Guardian: on Ethereum at block 26294960,
RocketStoragereports the guardian as0x0ccf14983364a7735d369879603930afe10df21e. That address has no deployed code in the on-chain read, so it is a single key rather than a contract. The privilege read lists it as the holder ofsetGuardian(address)andsetDeployedStatus()onRocketStorage, both with a fastest delay of 0 seconds and effect "config" . These let the key hand its role to another address and change the storage deployed flag; the sources read do not show that they move funds or replace contracts. - Bootstrap powers: a 2022 Sigma Prime review said a compromised guardian could "immediately spend all funds managed by the protocol DAO treasury" . The same review said a 2-of-3 multisig was planned . A later Saturn 1 review states the bootstrap mechanism is disabled in the deployed version and for the Protocol module . The on-chain read cannot show which guard functions the guardian still holds; the privilege parser marked the upgrade guards
multiple_guardsandmodifier_not_understood. - Security Council composition: per the team's response in the January 2026 Cantina review, "Currently, there is a single SC member. But this member is a multisig comprising 3 of the team members." Its threshold and signer identities are not in the sources, and the team says a veto needs 2 of the 3 keys: "So it does currently require 2/3 leaked keys from the multisig to veto an upgrade." The Houston docs said the guardian would start as the only SC member .
- Web check: a forum post (source reliability: community publication) says the team-controlled guardian was to be sole Security Council member until an appointment process was agreed . Etherscan labels the guardian address "Rocket Pool: Deployer 2" . A web search for the current council addresses, threshold and signers found nothing, so those stay unconfirmed.
- Veto risk (disputed): the auditors say that with
upgradeveto.quorumat 0.33 a single compromised SC key could veto upgrades . That audit is a copy hosted on the Rocket Pool site with unverified publication provenance. The same finding appears in the Cantina PDF atsources/docs/download-722.md. The team's answer is that the SC member is a 3-key multisig, so one leaked key is not enough, and that pDAO can replace members. Cantina marked the finding acknowledged. - Polygon rETH proxy: on Polygon (block 96353377) the rETH token proxy
0x0266f4f08d82372cf0fcbccc0ff74309089c74d1has proxy owner0x355b8e02e7f5301e6fac9b7cac1d6d9c86c0343f, aMultiSigWalletWithDailyLimitwithrequired: 1. The privilege read listsupdateImplementation(address)andtransferProxyOwnership(address)behindonlyProxyOwner, with the guard unresolved andfastest_delay_seconds: null, so the delay is unknown rather than zero. The scope is the Polygon rETH token only. DefiLlama lists Ethereum only for this protocol, so this read does not set a factor level. - Other multisigs: the on-chain read lists two 4-of-7 Safes labelled GMC and IMC in the docs. The docs list them as
GMCandIMC. Signers appear only as addresses with no identities, and whether these Safes hold any upgrade authority was not checked; the pDAO docs describe management committees as recurring treasury payees . - Benefit: a fast-acting small SC can stop a bad upgrade or switch off a faulty feature quickly. The team says SC powers "only [have] the power stop things from happening and even then only temporarily."
Upgradability
Upgrades go through the protocol's own registry pattern, not a standard proxy admin.
- Pattern: contracts register in
RocketStorageand are replaced byRocketDAONodeTrustedUpgrade. The on-chain read found no standard proxyadminorimplementationslots for the core contracts, which means the upgrade path is the registry (a missing slot is not proof of immutability). - Who: the oDAO proposes upgrades; the SC can veto;
proposalUpgradeis limited to executing contracts. The privilege read lists thebootstrapUpgradeguard asonlyGuardian + onlyBootstrapMode, withfastest_delay_seconds: null, meaning the delay is unknown rather than zero. - Delay: the official process page lists a 1-week upgrade delay after an oDAO vote of up to 2 weeks . Cantina describes
upgrade.delayas starting at 1 week (RPIP-60). The on-chain value was not read. Guardrails were added after the review: "upgradeveto.quorum >= 33% <= 100% upgrade.delay >= 1 day <= 30 days" . The 1-day figure is the permitted minimum, not the current delay. - Core invariants: an upgrade can replace any registered contract, including those that hold ETH and mint rETH, so the delay and veto are the main protection. The minipool manager can be replaced by governance .
- Exit window: Cantina noted validator exit queues of "~15 days" (an early-2026 observation, not a current measurement) against a 7-day delay and recommended 14 days; the team kept 7 days . For rETH holders, burning rETH depends on ETH liquidity rather than the validator queue: "burn can only happen if balance is covered by RETH contract balance and DepositPool" . Selling on a market is the alternative. A web check on 29 October 2026 found a beaconcha.in capture (source reliability: unverified publisher) showing , while an article of 10 October 2026 (source reliability: unverified publisher) reported The exit window therefore varies between about 1 day and more than 2 weeks across these sources, and neither source sets a level.
Parameter Controls
pDAO controls protocol parameters through on-chain votes, with partial bounds.
- Who: pDAO votes change settings and treasury spending; the SC can change flagged settings without delay; the oDAO controls its own settings and oracle submissions.
- Bounds: the pDAO-controlled
reduced.bondsetting had no guardrails and could be set "from zero to an arbitrarily large one" . The team's reply in the same review reads: "Fixed in commit 0e647e29. We have implemented the following guardrails per your recommendation: reduced.bond >= 1 ETH <= 4 ETH." That is a reviewed commit; the deployed configuration was not confirmed. - User harm: before the fix, a zero
reduced.bondwould have given 100% LTV, which the auditor calls risky for protocol security . With the stated 1 to 4 ETH bounds this is limited if the reviewed commit is what is deployed. The pDAO voting path (a proposal period, a challenge period and a veto vote) limits how quickly this can happen.
Decentralization Assessment
Rocket Pool sits in the middle of the spectrum: on-chain DAOs decide most things, but the safety layer is team-run.
- Roadmap: Houston moved pDAO execution from the core team to on-chain voting . The Security Council is to grow beyond the team: "The pDAO will elect additional members to the SC in the future."
- Key-person risk: the January 2026 Cantina review records the only SC member as a multisig of 3 team members ; the current council address and signers are not confirmed. The measured guardian address is a separate single key. Neither is shown to upgrade contracts or move user funds on its own, but the guardian's full remaining powers were not resolved. The documented worst cases are blocking upgrades and, if the guardian still holds legacy powers, bootstrap-era actions.
- Token distribution: voting power follows staked RPL under a square-root formula. Whether it matches stake concentration cannot be judged because voter shares were not measured.
Governance Risks
- Single points of failure: per the January 2026 Cantina review, compromise of two keys of the SC team multisig could block upgrades for weeks (the guardian key's current veto authority is not established), because replacing an SC member takes a pDAO vote ("could take several weeks" per Cantina ).
- Governance attack: a captured pDAO could set
upgrade.delayanywhere in the guardrail range of 1 to 30 days. - Voter apathy and concentration: the published quorum is 15% , but the top-10 share of votes cast is unmeasured, so apathy and delegate concentration remain an open question; a measured figure would settle it. A web search for recent pDAO and oDAO proposal outcomes and participation found no on-chain tallies. It found that Saturn 1 was on-chain proposal #27, raised 9 February 2026 . A community publication (source reliability: community publication) says the Saturn 2 process had progressed to snapshot voting , while another article (source reliability: unverified publisher) calls those votes onchain . This conflict was not resolved, and no Oracle DAO proposal outcomes were found.
- Centralization creep: no source read documents an accumulation of admin powers, and no dated proposal history could be retrieved to test it. The official process page says settings, treasury spends and council membership are executed by the Protocol DAO .
- Oracle DAO majority: a 2021 audit claim says a malicious majority of oracle nodes could take over the system and pay out arbitrary prices or tokens . Whether this still applies to the current deployment is unconfirmed. The official upgrade path today has an oDAO majority vote, a 1-week delay and a Security Council veto .
Data Completeness
- Available: the on-chain control read of Ethereum at block 26294960 (partial status), audit reports and review PDFs on the Houston and Saturn 1 designs (Sigma Prime 2022, Bailsec, Cantina January 2026), and docs on pDAO, oDAO and Security Council.
- Missing: the current Security Council multisig address, threshold and signers; the guardian's remaining powers; measured voter concentration (top-10 share of votes cast); the on-chain delay values; a current dated exit window; and recent proposal outcomes and participation. The Optimism, zkSync and Scroll reads were not made (not read or forbidden per the control file). The Polygon rETH proxy is covered under Admin Key Analysis.
- Provenance: several audit copies have unverified publication hosts; they inform confidence only and do not lower the assessment of the protocol.
Score: 3.5/10
Rubric security-score-v2.5; data completeness 0.00
- Admin control
- 3.50 points
- Change delay
- 3.50 points
- Decision process
- 3.50 points
| Factor | Level | Evidence | Weight | Points |
|---|---|---|---|---|
| Admin control | — | not disclosed | 0.45 | 3.50 |
| Change delay | — | not disclosed | 0.30 | 3.50 |
| Decision process | — | not disclosed | 0.25 | 3.50 |
56 citations · click markers to view evidence
Cross-Analysis
Cross-Risk Analysis: Rocket Pool
Rocket Pool's biggest combined risk is not a single bug but that rETH (the liquid staking receipt token, about 362,000 ETH of supply value inside USD 1.12bn of total protocol TVL) can be exited directly only through a buffer of about 0.73% of supply. Larger exits depend on DEX liquidity, whose depth was not measured, or on validator exits, for which no maximum delay is stated. The protocol also relies on a ten-member oracle committee and on Saturn 1 contracts (live since 18 February 2026) whose audit-to-deployment link is unproven. The reports agree the staking mechanics and team identity are strong, but the governance and admin-key picture is incomplete, so the governance score of 3.5/10 reflects missing data more than a measured weakness. Confidence is moderate: the only recorded loss is a 2023 USD 3.8M phishing entry whose scope is unconfirmed, but several protections (rate-change limit, upgrade delay, deployed-code match) were not read on-chain.
Scope: the products and tokens covered are rETH (the liquid staking receipt token), RPL (the governance and collateral token), and node operators running minipools or the newer megapools . Likelihood bands below are qualitative assessments, not annual probabilities.
Top Risks for a Holder
-
rETH exit discount under stress (rETH; RPL is not directly affected).
- What can go wrong: a wave of redemptions exceeds the direct-redemption buffer, so holders who still want to exit sell on DEXs (depth not measured) or wait for liquidity or validator exits (no maximum delay stated). Assessment: if sellers outnumber available liquidity, rETH could trade below its protocol rate, and the loss would be the discount for holders who sell at it.
- Evidence: the rETH contract and deposit pool hold about 2,646 ETH of collateral, a collateral rate of 0.73% . Supply is about 362,000 ETH . DEX depth was not measured . A community forum post (source reliability: community publication) says mainnet rETH liquidity fell from over USD 30M to about USD 2.5M, an unconfirmed lead . A captured Balancer page (source reliability: unverified publisher) says most pools move to withdrawals only on 30 October 2026, the day after this analysis; this is also an unconfirmed lead . DeFiLlama shows a 7-day TVL change of -11.7% .
- Magnitude: the buffer covers about 0.73% of rETH supply, which is worth about 362,000 ETH; total protocol TVL is about USD 1.12bn. The discount size is unknown because depth is unmeasured. No rETH depeg history exists to calibrate it .
- Likelihood: medium (assessment) that a visible discount appears in a stress event, because the buffer covers 0.73% of supply and DEX depth is unmeasured. A measured 2% price-impact depth across venues would move this band up or down.
-
Loss of staker ETH through Saturn 1 megapool or deposit-path bugs, or an unfunded loss above the 4 ETH operator bond (rETH, megapool node operators).
- What can go wrong: the newest code is the least time-tested and carries many open items. A bug, or a validator loss larger than the operator's bond, could cost staker ETH. Who absorbs a loss above the bond is not confirmed, and no loss-cover fund was found.
- Evidence: Bailsec reported 144 findings including 26 High and cannot "state with high confidence that no critical issues remain" . Of 458 distinct audit findings, 235 are fixed, and 52 critical, high or medium findings are not marked fixed . Deployed code was never compared with any audited version . The docs do not say who absorbs a loss above the 4 ETH bond , and no insurance or reserve fund was found . The Immunefi bounty maxes at USD 150,000, about 0.013% of TVL .
- Magnitude: the share of TVL sitting in megapools was not measured, so a USD figure cannot be given. The worst case is a fund-draining bug in contracts that custody staker ETH (severe). Who bears a bond-exceeding loss, and how fast it would show in the rETH rate, depends on the Saturn 1 loss waterfall, which was not found.
- Likelihood: low to medium. Four independent firms reviewed Saturn 1 and both Cantina criticals are marked fixed. Against that, the code is under nine months live, and the repository check still finds code for several acknowledged Saturn items. The fact that would decide it: a verified-source comparison of the live megapool contracts against the audited commits.
-
Trusted-key control of the rETH rate and of contract upgrades, and governance-adjustable RPL staking rewards (rETH, RPL stakers, node operators).
- What can go wrong: a colluding oracle majority misreports the rETH rate, or a compromised oDAO and Security Council allow a malicious upgrade of contracts that hold ETH. Separately, RPL stakers and node operators face reward changes, because staker rewards depend on governance-adjustable splits and on RPL inflation planned to phase out with Saturn 2.
- Evidence: the oDAO has 10 members and the rate is pushed by it; in practice 6 of 10 must agree . The only documented rate limit is a governance specification, not a confirmed live setting . An upgrade can replace any registered contract, including those holding ETH . The guardian is a single key with zero-delay
setGuardian, and its remaining powers were not resolved. RPL staker rewards depend on governance-adjustable splits and on RPL inflation planned to phase out with Saturn 2 . - Magnitude: a rate error of up to 2% per update (if the specification is live and the supply fix is effective) applies to the whole rETH supply, about 7,200 ETH per 2% of the roughly 362,000 ETH supply value (calculation from the cited supply and rate). A malicious upgrade could in principle reach the whole TVL. For RPL stakers and operators, total RPL issuance is about 5% of supply a year (calculated from the on-chain interval rate) . Operators receive the majority of it, but their exact share is not cited, so the exposed reward stream is smaller than total issuance and its size is a gap .
- Likelihood: low for the key-control events. It needs a majority of the ten oracle nodes, or the holders of the upgrade and veto powers, to be compromised or to collude, and no such event is recorded. Assessment: a change to RPL staker rewards is likely over time, because the documented plan phases out the inflation rewards, but the timing and the size of the split changes are not stated. The missing facts that would change the band are the on-chain upgrade delay, the live rate-change limit and the guardian's remaining powers.
Combined Attack Vectors
Vector 1: Oracle rate error meets a thin redemption buffer and lending collateral
- Scenario:
- A majority of the 10 oDAO nodes (6) submits a wrong rETH balance or supply figure. Sigma Prime found that the ETH change was limited but rETH supply was not, and that supply also influences the price .
- The rate changes. Each step is bounded only if the live per-update limit (unread; 2% in the specification) is set and the supply fix is effective. The Sigma Prime fix is marked resolved, but deployed code was not compared with any audited version .
- Assessment: if the rate is too high, direct redemptions at that rate would draw on the roughly 2,646 ETH buffer. The effect on DEX prices was not measured.
- Lending markets that price rETH from the protocol rate could misprice it. The largest holder listed is an Aave rETH contract at 11.3097% of supply (source reliability: unverified publisher; a lead, not a verified measure) . How that market prices rETH and how much exposure it has were not checked.
- Domains: economic (oracle design, redemption buffer), governance (oDAO trust), code (a Sigma Prime finding that the supply limit was missing, marked resolved).
- Likelihood: low. It needs majority collusion; no event is recorded. If a per-step cap of 2% is live and the supply fix is effective, repeated updates would be needed to exceed it.
- Impact: severe. Assessment: a confirmed cap would bound each step but not the number of steps. Without a confirmed cap and an effective supply constraint, the size of one step is unknown, and the 0.73% buffer is small against a run on a mispriced rate.
Vector 2: Megapool exit-accounting failure with no loss-cover fund
- Scenario:
- A node operator fails to notify an exit.
- The oDAO also fails to challenge the megapool before funds arrive. A call to
distributethen treats principal as rewards . The team calls this intended behaviour that relies on the challenge process. - No reserve fund was found, and the loss rule above the 4 ETH bond is unconfirmed, so who bears any residual is not established.
- Domains: economic (exit accounting, no loss waterfall, no insurance) and governance (the oDAO must challenge in time). No finding from the code report is shown to cause, amplify or limit this path.
- Likelihood: low. No realised shortfall is recorded, and the stated protection is the oDAO challenge, which also fines the operator .
- Impact: moderate to severe. The size of a single event was not measured, and who would bear it depends on the unconfirmed loss rule above the 4 ETH bond.
Vector 3: Upgrade-path capture through a thin veto layer
- Scenario:
- An oDAO majority proposes a malicious contract replacement. A Bailsec finding says there is no expiry on upgrade proposals, so a stale proposal can be executed later (acknowledged; the code is still present in the repository check).
- The documented delay is 1 week, but the live
upgrade.delaywas not read, and the guardrail permits a minimum of 1 day . A Cantina Medium (acknowledged) says the veto quorum and delay lacked guardrails. - The only Security Council member read on-chain is a contract with threshold 2 and three owner addresses . The owners were not checked for being keys or contracts, and the sources do not say who they are . Separately, the team stated in January 2026 that the member was a multisig of three team members . The sources do not show that the current owners are those team members.
- Assessment: if the veto is not used, or two owners cannot or do not act, a malicious proposal would not be stopped by the council. That is a separate case from a malicious veto, where two owners block a legitimate fix; replacing a council member needs a pDAO vote that "could take several weeks".
- A malicious upgrade then swaps in a contract that controls ETH or the rate.
- Exit during the notice window: the delay only helps holders if they can leave. Holders beyond the 0.73% direct-redemption buffer would need DEX liquidity (depth not measured) or validator exits. Assessment: the delay gives notice, and liquidity decides whether the notice is usable.
- Domains: code (upgrade-expiry and guardrail findings, upgradeable-contract finding), governance (delay, veto, key concentration), social (the team is named, but the owners behind the threshold-2 contract are not identified).
- Likelihood: low. It needs a compromised oDAO majority and a failed veto; the one-week design and public RPIP process raise the cost. Reading the on-chain delay would sharpen this.
- Impact: catastrophic in the worst case, because the registry pattern lets an upgrade replace the contracts holding ETH.
Vector 4: Exploit news meets an exit with no global pause
- Scenario:
- A Saturn 1 bug becomes known, or an exploit begins.
- The Security Council can act instantly only with a "strictly whitelisted set of pause-type powers" . Recent examples are disabling solo migration, node deposits and node registrations, and the sources describe no global pause.
- Some holders sell on DEXs (depth not measured) while the buffer holds under 1% of supply (see Risk 1).
- Assessment: rETH could trade at a discount even if the exploit is contained.
- Domains: code (new-code risk), governance (limited emergency powers), economic (redemption liquidity).
- Likelihood: low to medium. It depends on a bug surfacing, and the size of any discount depends on DEX depth, which was not measured.
- Impact: severe for rETH sellers; the loss is realised price, not necessarily lost backing.
Systemic Risk Factors
- Dependencies: the dependency graph shows 100% of TVL backed by WETH, a node it marks unresolved, with no scored protocol dependency and
warnings: [](TVL date 2026-10-29) . Ethereum's own validator rules (slashing, inactivity penalties, exit queue) are the real backing risk and are not a scored dependency. The economic dependency assessment owns this route. - Graph coverage: the graph assigns 100% of backing to the unresolved WETH node and 100% to Ethereum, with composition dated 2026-10-28 and TVL dated 2026-10-29 . The empty warnings list reflects limited graph coverage, not proof of no contagion. No external vault, lending market or yield token is established as a backing layer, so any later-identified intermediary goes to the economic dependency assessment. User-position leverage and RPL reflexivity belong to yield design, and residual validator losses to solvency and redemption.
- Downstream holders: the graph lists no downstream holders as warnings. The Etherscan page (source reliability: unverified publisher) shows Aave's rETH contract at 11.3097% of supply and the top 10 holders at 29.66% . Assessment: correlated liquidation or withdrawal by those holders could hit the same exit venues, whose depth was not measured.
- DEX venue concentration: the reported Balancer wind-down date of 30 October 2026 (source reliability: unverified publisher) is the day after this analysis, and the report stays unconfirmed. The docs name Balancer as the recommended venue . A Balancer v3 pool of about USD 4.3M was captured (source reliability: unverified publisher); pool snapshot dates were not confirmed.
- Stale price feed: the Maker rETH/USD feed looks dormant since about March 2025, per the on-chain read; integrators that use it get an old price .
- RPL holders: RPL supply grows about 5% a year at the measured inflation rate (a calculation from the on-chain interval rate), and staker rewards depend on governance-adjustable splits and on inflation planned to phase out with Saturn 2 . An RPL price fall weakens secondary collateral and operator incentives but leaves rETH backed by staked ETH .
- Chain level: all counted TVL is on Ethereum, so there is no L2 sequencer or bridge risk on TVL. Bridged rETH and RPL exist on L2s and carry no counted TVL. The Polygon rETH proxy is owned by a
MultiSigWalletWithDailyLimitwithrequired: 1, a single-signer risk for Polygon holders only. - Macro: an ETH price drop or a staking-yield compression lowers both TVL and operator incentives. The 7-day TVL decline of 11.7% is one capture and does not separate price from outflows.
- Correlation: Assessment: a rETH discount and an oracle error could both affect the same exit venues and the same oDAO signal.
Risk Amplifiers
- Thin buffer + unmeasured DEX depth + reported venue closure (economic, social): Assessment: if the Balancer wind-down and the forum liquidity drop are accurate, a panic could move the price more than usual. Both are unconfirmed leads from weak sources.
- New code + unlinked audits + small bounty (code, economic): 26 High Bailsec findings, no audit-to-deployment comparison and a USD 150,000 cap on critical rewards leave a hacker with a far larger payoff than the bounty pays. Assessment: this is an incentive gap, not a documented exploit.
- Rate trust + composability (economic, governance): rETH is collateral elsewhere, so a wrong rate is not contained inside Rocket Pool.
- Single-key guardian + unresolved guard parsing (code, governance):
setGuardianhas zero delay, and the privilege parser could not resolve several upgrade and fund-moving guards, so the true worst case of a guardian compromise is unknown. - Team-run veto + named team (social, governance): the team stated in January 2026 that the council member is a multisig of three team members. The on-chain read shows threshold 2 and three unidentified owners. Assessment: if the owners are those team members, the veto depends on two named people acting.
- Wrong deposit-pool address in reads (overview, code): see Contradictions. If the privilege read examined a superseded contract, the control picture for the live pool is incomplete.
Risk Mitigators
- Operator first loss: operators bond 4 ETH per validator and take losses first up to that amount . This is thinner than the older 8 and 16 ETH bonds.
- Review depth: four independent firms reviewed Saturn 1 within January and February 2026, and the audit-depth score is the strongest code factor (7.50). Hashlock's report is on the auditor's own site; the Cantina, Bailsec and Sigma Prime Saturn copies are project-hosted with unverified provenance, which lowers confidence, not the credit.
- Hashlock scope: Hashlock's January 2026 report lists megapool, deposit, node, token and network contract paths in scope, and its scope review agrees with the audit index. Its one Medium finding, on RPL slashing, is marked resolved. This credit applies to those paths, not to every earlier minipool issue .
- Fixed critical findings: 9 of 11 critical groups are fixed. The two unfixed ones are from 2021 and concern deployment-phase powers, and the registry now reports
getDeployedStatus: true, which closes the settings window only. - Governance design: delay plus Security Council veto plus public RPIP proposals, and bootstrap mode disabled on both DAOs .
- Organic yield: rETH return comes from Beacon Chain, fee and MEV rewards, not emissions, so there is no reflexive emission loop for rETH.
- Operating history: live on mainnet since 9 November 2021 with several upgrades delivered; the only recorded incident (2023, USD 3.8M phishing) appears in weak sources to be a single holder's wallet, not the contracts, though no first-party statement confirms this.
- Rate-change controls: the specification caps each rETH rate update, and a Sigma Prime finding on the missing supply limit is marked resolved. Both depend on the live setting, which was not read, and the fix has not been compared with deployed code.
- Third-party scores: CertiK Skynet 77.42 (BBB) and Exponential risk score 9.0. These do not test the redemption buffer or oDAO trust, so they do not offset Risks 1 and 3.
Contradictions & Inconsistencies
- Security Council signers. Governance says the council's threshold and signers are "not in the sources" and calls them unconfirmed, while the overview read a 2-threshold contract with 3 owners on-chain at block 26296425. The team's January Cantina reply says the member is a multisig of three team members; the on-chain read does not show whether the three owners are those people, or whether they are keys or contracts.
- Governance score vs evidence. Governance reports all three factors as "not disclosed", with data completeness 0.00 and a 5.0 cap, yet the report itself contains a substantial documented structure. The 3.5/10 reflects unread delay values and unresolved guard parsing, not a measured weakness. It should not be read as "worse than peers".
- Which deposit pool is live. The social report lists the
Depositcontract at0xDD3f…5AF8as the current core contract. The overview shows a registry lookup returning0xCE15…4fdCfor the live deposit pool, and notes the privilege read examined the other address. Fund-moving guard conclusions for the live pool are therefore not established. - Redemption buffer size. The Saturn 1 docs describe a 1% buffer, Sigma Prime refers to a 10% buffer, and the measured collateral rate is 0.73%. The sources do not reconcile these.
- Oracle committee size. Exponential says consensus needs 9 of 16 oDAO members . The on-chain read shows 10 members and a quorum of 5.1 votes, so the Exponential text looks outdated. Exponential also says there are no documented hacks, while DeFiLlama lists the 2023 phishing record.
- Upgrade delay. Governance cites a 1-week delay and the team's "2 weeks" minimum; Exponential states a timelock of at least 48 hours; the on-chain delay was not read. These are not reconciled.
- Pre- vs post-Saturn loss rule. The economic report's cited FAQ describes spreading losses across rETH under 8 and 16 ETH bonds. The current docs say the operator's share absorbs losses but are silent above 4 ETH.
- Exit queue. Governance records a validator exit window from about 1 day to more than 16 days across weak sources. Economic security treats direct rETH redemption as depending on pool liquidity, not the validator queue.
- Audit counts. The overview counts 18 audit documents and code security counts 16 included reports, because two duplicate copies were excluded. This is a counting difference, not a conflict.
- Pause history. The economic report says no exit halt or pause is reported, while the governance and social reports cite recent Security Council pauses of solo migration, node deposits and node registrations without dates. Whether any pause affected exits is not established.
Overall Risk Profile
Rocket Pool is a mature, publicly staffed liquid staking protocol with organic yield and deep audit history. Its composite risk is moderate. The combination that matters is a thin direct-exit buffer (0.73%) with unmeasured DEX depth and an unconfirmed report of a closing venue, a newly launched upgrade whose live code was never matched to audits, and trusted-committee control of the rate and upgrades. A possible failure mode is a rETH discount to its protocol rate during a stress or exploit scare, which would cost sellers the discount but not backing. The worst case is a bug or malicious upgrade in the Saturn 1 or registry layer that moves staker ETH, where the USD 150,000 bounty, the absence of a found loss-cover fund and a veto whose holders are not identified on-chain leave little buffer. No peer data is in the sources, so no ranking against other liquid staking protocols is made; the protocol's own record shows deep audit history and a named team.
Data Gaps
-
Some audit reports were unavailable or could not be read in full.
-
Some protocol documentation could not be read because the document contains images without readable text.
-
Some source material is unavailable because it is not published.
-
An audit report is unavailable because it could not be collected.
-
Collection of linked source pages stopped at the collection limit; some material may be missing.
-
Collection of GitHub repositories stopped at the collection limit; some material may be missing.
-
Current deployment addresses could not be confirmed for some audited contracts.
-
Source summaries cover only the first 80000 characters of 11 document(s); later content was not assessed.
-
DEX depth. The 2% price-impact depth for rETH across venues was not measured; captured pool pages give only total value and have unconfirmed dates. This limits sizing of Risk 1 and Vectors 1 and 4.
-
Megapool share of TVL. Not measured, so Risk 2 and Vector 2 cannot be sized in USD.
-
Audit-to-deployment link. 70 audit-named contracts have no attributed live address, and no deployed code was compared. Post-January 2026 fixes and Bailsec's re-review status are not stated.
-
Live on-chain parameters. The rETH delta limit, the deposit delay,
upgrade.delayand upgrade-function delays were not read, so Vectors 1 and 3 rest on documented, not confirmed, values. -
Guardian powers and council owners. The guardian's remaining powers and the identities of the three Security Council owners are unknown. Vote concentration on the pDAO was not measured.
-
Guardian observation. The code report's one independent observation, that the guardian is a single key, is published as a hypothesis (informational) and is not promoted here . The audit index could not confirm the scope of the Sigma Prime Saturn review or the Bailsec report.
-
Deposit pool read. The privilege read examined a different contract than the live registry address, so fund-moving controls on the live pool are not confirmed.
-
Loss waterfall. The Saturn 1 rule for losses above the 4 ETH bond was not found.
-
Other chains. Optimism, zkSync and Scroll were not read. Bridge setups for bridged rETH and RPL were not assessed, and they carry no counted TVL.
-
Weak sources. The holder concentration, pool values, Balancer wind-down date, forum liquidity figure and phishing attribution all come from community or unverified publications and remain unconfirmed leads.
47 citations · click markers to view evidence