Worked example · 2023-06
Chibi Finance panic / onlyGov residual-authority exit-scam — Arbitrum — 2023-06-27
Summary
Chibi Finance was a DeFi yield-aggregator on Arbitrum that publicly marketed itself as a community-governance protocol providing automatic rewards to users who staked tokens. The protocol's smart-contract architecture distributed authority across eight separate contracts, each of which exposed a panic() function gated by an onlyGov modifier — meaning the function was callable by an admin contract but not by ordinary users. The panic() function was structured as an emergency-withdrawal pathway: it withdrew all tokens from the contract's pool and sent them to a designated address.
On 2023-06-27, the Chibi Finance deployer EOA executed the rug:
- Transferred admin rights for the eight Chibi Finance contracts to a malicious contract under operator control. This step was legitimate at the contract layer — the deployer held the authority to assign the
_govrole. - The malicious
_govcontract invokedpanic()across all eight Chibi protocol contracts in rapid succession, executing emergency withdrawal of all funds. - The drained tokens — approximately $1M aggregate value — were swapped into ETH (~555 ETH) and bridged from Arbitrum to Ethereum.
- The bridged ETH was routed through Tornado Cash, closing the on-chain recovery window structurally.
The CHIBI governance token's price fell ~98% as the news propagated; per Halborn's same-day analysis ("Explained: The Chibi Finance Rug Pull (June 2023)"), the rug was the 12th Arbitrum-based protocol to exit-scam users in 2023, with the cohort cumulatively losing >$14M by mid-year.
The case is OAK's canonical T1.003 example for the onlyGov residual-authority sub-class: the residual authority surface was documented in the verified contract source as an onlyGov modifier on a panic-style emergency function, and the deployer's authority to assign _gov was the load-bearing operator-control surface. The exploit was therefore not a bug — it was the intended operator-control pathway, dressed in community-governance marketing. Halborn, CertiK, Cointelegraph, CoinDesk, CryptoSlate, BeInCrypto, CryptoPotato, Tokenist, Technext, BSCN, and CoinJournal contemporaneous coverage triangulates the technical root-cause to the onlyGov panic-function pathway.
Why this is structurally significant
T1.003 has at least three structurally distinct 2023 sub-classes that share the Technique number but diverge on detection and mitigation:
Router-rerouting via retained authority — Hope Finance (
/examples/2023-02-hopefinance.md). The retained authority is on a router contract; the failure mode is silent token-flow redirection at runtime.onlyGovpanic-function residual authority — Chibi Finance (this case). The retained authority is the deployer's ability to assign the_govrole; thepanic-function is documented in the verified source. The failure mode is overt emergency-withdrawal extraction across multiple contracts.Oracle-modification via residual oracle-admin authority + cross-chain deployer cluster — Magnate Finance (
/examples/2023-08-magnate-finance-base-deployer-cluster.md). The retained authority is on the lending-protocol price oracle; the failure mode is collateral-price inflation followed by lending-pool drainage; M04 cluster-attribution is the canonical defender-tooling primitive.
The three sub-classes share the structural property that the residual authority surface was checkable on-chain at the contract layer before the rug, but they require different detection methodologies:
- Hope: bytecode review + proxy-implementation analysis on Arbiscan.
- Chibi: verified-source review + role-assignment privilege graph (M05 authority-graph enumeration).
- Magnate: M04 funder-graph clustering + cross-chain deployer-EOA correlation.
Chibi's structural distinguishing feature is that the residual authority was named in the source code and visible in the verified Etherscan / Arbiscan contract. The defender did not need to recompile bytecode (Hope) or trace cross-chain deployer linkage (Magnate); the panic() function gated by onlyGov was readable in plain Solidity. The case is therefore a particularly clean teaching example for the source-level authority-graph enumeration mitigation pattern (OAK-M05).
Three structural features distinguish the case within the T1.003 cohort:
The residual authority is multi-contract. Eight separate Chibi Finance contracts each exposed a
panic()function gated by the sameonlyGovmodifier. The deployer's single_govreassignment unlocked thepanicpathway across all eight contracts simultaneously. T1.003 detection methodology should enumerate the residual-authority surface across the protocol's full contract graph, not just on the headline contract.The Tornado Cash recovery-window-closure is the standard 2023 endpoint. Chibi's proceeds reached Tornado Cash within hours of the rug; standard recovery-window-closure profile across the 2023 T1.003 cohort. The structural property is shared with Hope Finance, Magnate Finance, and the broader 2023 rug-pull cohort whose proceeds reach a public mixer in the same transaction window.
The Arbitrum-2023 cohort scale. Chibi was the 12th Arbitrum-based exit-scam in 2023; the cohort cumulatively lost >$14M by mid-year. The chain-distribution is a gas-economics signal (cheap deployment + active retail capital + young listing-compliance regimes); future T1.003 examples should record the chain-cohort context.
Timeline (UTC)
| When | Event | OAK ref |
|---|---|---|
| Pre-2023-06 | Chibi Finance launches on Arbitrum as DeFi yield-aggregator; publicly markets community-governance framing; deploys eight protocol contracts each gated by onlyGov panic() modifier; _gov role assignable by deployer |
(T1.003 standing surface) |
| 2023-06-27 | Deployer transfers _gov admin rights for the eight Chibi Finance contracts to a malicious contract under operator control |
T1.003 setup (residual-authority-reassignment) |
| 2023-06-27 (same window) | Malicious _gov contract invokes panic() across all eight Chibi protocol contracts in rapid succession; emergency-withdrawal extracts ~$1M aggregate |
T1.003 + T5.001 execution |
| 2023-06-27 (within hours) | Drained tokens swapped into ~555 ETH; bridged from Arbitrum to Ethereum | T5 outflow |
| 2023-06-27 (within hours) | CHIBI token falls ~98% as news propagates; CertiK, Halborn, Cointelegraph, CoinDesk publish forensic coverage | (defender response / market signal) |
| 2023-06-27 onward | Bridged ETH routed through Tornado Cash; on-chain trail effectively foreclosed at the mixer boundary | T7.001 |
| Continuing | No public recovery; no civil-forfeiture action; operators remain pseudonymous | (recovery state) |
What defenders observed
- Pre-event (verified-source layer): the
panic()function gated byonlyGovwas named in the verified contract source on Arbiscan. A standard source-level review (OAK-M05 authority-graph enumeration) would have surfaced the residual authority across the eight protocol contracts. Defender lesson: source-level review of every authority-bearing modifier across the protocol's full contract graph is the canonical 2023 mitigation pattern; the residual authority was hiding in plain sight, not behind proxy indirection or bytecode obfuscation. - At-event (role-assignment signal): the deployer's transfer of
_govto the malicious contract is a single transaction event observable on Arbiscan. A protocol-monitoring indexer subscribed to_govrole-assignment events would have detected the precondition; the malicious_govcontract's subsequentpanic()calls are the extraction signal. Vendor coverage of this monitoring pattern at v0.1 is partial; v0.x mitigation work should formalise the integration pattern at the wallet / venue UX layer. - At-event (CHIBI-price signal): the governance token fell ~98% as the news propagated; a price-surface indexer integrated with on-chain monitoring would have flagged the rug within minutes. The price signal is secondary to the on-chain extraction signal but operationally useful for cohort-scale alerting.
- Post-event: standard Tornado-Cash recovery-window-closure profile. Within hours of the rug, proceeds had reached the mixer and the on-chain trail was effectively foreclosed at the mixer boundary. The base-rate expectation for T1.003 / T5.001 incidents whose proceeds reach a mixer within the same operational window is total loss with no recovery; Chibi is consistent with the base rate.
What this example tells contributors writing future Technique pages
- T1.003 has at least three structurally distinct 2023 sub-classes. Hope Finance (router-rerouting), Chibi Finance (this case,
onlyGovresidual authority), Magnate Finance (oracle-modification + cross-chain cluster). The three together cover the canonical 2023 T1.003 reference triple. T1.003 contributors should preserve the sub-class distinction; the detection methodology diverges across the three. - Multi-contract authority-graph enumeration is the load-bearing M05 pattern. Chibi's residual authority spanned eight protocol contracts; a single-contract review would have missed the full extraction surface. M05 enumeration must cover the protocol's full contract graph, not just the headline contract.
- Source-level review of
onlyGov/onlyAdmin/onlyEmergencymodifiers is the canonical 2023 defender practice. Chibi'spanic()function was named in the verified source; the residual authority was not hiding behind proxy indirection. Defender practice should treat every authority-bearing modifier across the protocol's full contract graph as a residual-authority surface until proven gated by timelocked governance. - The Tornado Cash recovery-window-closure is the standard 2023 endpoint. Future T1.003 examples should record whether proceeds reached a public mixer within the same operational window; if so, the recovery-window is structurally foreclosed and the defender intervention surface narrows to forward-going user exposure.
- Cohort-scale chain-distribution is a gas-economics signal. Chibi was the 12th Arbitrum-based exit-scam in 2023; cohort-scale chain-distribution is informative for defender-tooling prioritisation. Future T1.003 examples should record the chain-cohort context.
Public references
[halbornchibi2023](proposed) — Halborn, "Explained: The Chibi Finance Rug Pull (June 2023)": https://www.halborn.com/blog/post/explained-the-chibi-finance-rug-pull-june-2023[certikchibi2023](proposed) — CertiK, "Chibi Finance Incident Analysis": https://www.certik.com/resources/blog/chibi-finance-incident-analysis[cointelegraphchibi2023](proposed) — Cointelegraph, "Chibi Finance $1M alleged rug pull: How it happened": https://cointelegraph.com/news/chibi-finance-1-million-alleged-rug-pull-how-it-happened[coindeskchibi2023](proposed) — CoinDesk, "Chibi Finance Rug Pulls Users for $1M, CHIBI Falls 98%": https://www.coindesk.com/tech/2023/06/27/chibi-finance-rug-pulls-users-for-1m-chibi-falls-98[cryptoslatechibi2023](proposed) — CryptoSlate, "Chibi Finance becomes 12th Arbitrum-based protocol to rug users in 2023": https://cryptoslate.com/chibi-finance-becomes-12th-arbitrum-based-protocol-to-rug-users-in-2023/[cryptopotatochibi2023](proposed) — CryptoPotato, "Arbitrum-Based DeFi Project Chibi Finance Rug Pulled: Over $1 Million Drained": https://cryptopotato.com/arbitrum-based-defi-project-chibi-finance-rug-pulled-over-1-million-drained/[beincryptochibi2023](proposed) — BeInCrypto, "Chibi Finance Pulls the Rug: Users Lose Over $1M": https://beincrypto.com/chibi-finance-pulls-rug/[bscnewschibi2023](proposed) — BSCN, "Chibi Finance: Arbitrum-Based DeFi Project Allegedly Performs $1M Rug Pull": https://bsc.news/post/chibi-finance-arbitrum-based-defi-project-allegedly-performs-1m-rug-pull[tokenistchibi2023](proposed) — Tokenist, "Arbitrum-Based DeFi Project Chibi Finance Disappears With a Million Dollars": https://tokenist.com/arbitrum-based-defi-project-chibi-finance-disappears-with-a-million-dollars/[technextchibi2023](proposed) — Technext, "Chibi Finance Rug Pull; Calls for due diligence increase as crypto users lose over $1m": https://technext24.com/2023/06/29/chibi-finance-rug-pull-calls-for-due-diligence-increase-as-crypto-users-lose-over-1m/[coinjournalchibi2023](proposed) — Coin Journal, "Chibi Finance team executes $1M rug pull, CHIB plummets 98%": https://coinjournal.net/news/chibi-finance-allegedly-executes-1m-rug-pull-on-arbitrum-chib-plummets-98/[chainalysis2025rug]— broader cohort retrospective.[slowmist2024report]— broader 2024 cohort retrospective.
Discussion
Chibi Finance is OAK's canonical 2023 worked example for the onlyGov residual-authority sub-class of T1.003. The structural distinguishing feature is that the residual authority surface was named in the verified contract source — the panic() function gated by onlyGov is readable in plain Solidity on Arbiscan. The case is therefore a particularly clean teaching example for the source-level authority-graph enumeration mitigation pattern (OAK-M05): the residual authority was hiding in plain sight, not behind proxy indirection or bytecode obfuscation. Defenders performing standard source-level review of every authority-bearing modifier across the protocol's full contract graph would have surfaced the residual authority pre-rug.
The case structurally complements Hope Finance (/examples/2023-02-hopefinance.md — router-rerouting via retained authority) and Magnate Finance (/examples/2023-08-magnate-finance-base-deployer-cluster.md new — oracle-modification via residual oracle-admin authority + cross-chain deployer cluster). The three together cover the canonical 2023 T1.003 sub-pattern reference triple.
For OAK's broader cohort coverage at v0.1, the Chibi case is the cleanest single-chain (Arbitrum) T1.003 example with multi-source forensic triangulation (Halborn, CertiK, Cointelegraph, CoinDesk, CryptoSlate, BeInCrypto, BSCN, Tokenist, CryptoPotato, Coin Journal, Technext); meets the v0.1 standard for "verified across at least 2 independent forensic providers" with substantial margin. The case anchors the v0.1 T1.003 example set as the source-level-readable sub-class, distinct from the bytecode-review (Hope) and cluster-attribution (Magnate) sub-classes.