PancakeSwap
BSC-primary multi-version DEX: V2 (Uniswap V2 fork), V3 (Uniswap V3 fork with concentrated liquidity), StableSwap (Curve-style), and Infinity/V4 (hooks + singleton vault, launched April 2025). Also deployed on Ethereum, Aptos (Move language, non-EVM), Base, zkSync Era, opBNB, Arbitrum, and Linea. CAKE is the native governance/utility token bridged via LayerZero OFT. Perpetuals product powered by ApolloX/APX Finance. Governance via Snapshot (cakevote.eth) with team multisig execution; Tokenomics 3.0 (May 2025) retired veCAKE in favor of 1 CAKE = 1 vote.
DeploymentsBNB Chain · $1.7B
01
Risk profile at a glance
1 red · 5 yellow · 7 green 02
Categories & evidence
184 factors · 13 categoriesCode & audits Green 15 25 of 25
RD-F-009 red Formal verification coverage No formal verification (Certora, Kani, Halmos, or equivalent) detected for any PancakeSwap surface. V3 is a Uniswap V3 fork which itself has no formal verification. Infinity uses Foundry for testing but no FV specification files detected in repo inspection. At $1.73B TVL, critical invariants (k=xy for V2, tick-bound invariants for V3, vault balance invariants for Infinity) are assessable but unverified. RD-F-001 yellow Audit scope mismatch Six independent audit firms cover distinct PancakeSwap surfaces (SlowMist, PeckShield, BlockSec, Hexens, OtterSec, Zellic). Infinity/V4 has 3 external firm audits (Hexens, OtterSec, Zellic) conducted in 2024 prior to April 2025 launch; OtterSec resolved 2 high-impact vulnerabilities pre-launch. Commit-SHA-to-deployed-bytecode matching was not achievable via static read. V3 audits (March 2023) are ~37 months old. Gap risk exists for any post-audit code changes between final 2024 Infinity audit and April 2025 launch. RD-F-002 yellow Audit recency Infinity core/periphery: audited 2024, launched April 2025 — within 365 days. V3 core: audited March 2023 — approximately 37 months as of April 2026, exceeding the 24-month yellow/red threshold. V2: audited 2021–2022, >48 months. The dominant TVL surface (V3 on BSC) carries a 37-month-old audit making the blended score yellow. RD-F-005 yellow Audit firm tier OtterSec, Zellic, Hexens are established Tier-2 firms. SlowMist, PeckShield, BlockSec are Tier-2 BSC-specialised firms. No Tier-1 firm (Trail of Bits, OpenZeppelin, ConsenSys Diligence, Certora, Sigma Prime, Spearbit) has audited PancakeSwap. CertiK published a hooks security blog post, not a full engagement. Highest tier present is Tier-2. RD-F-006 yellow Audit-to-deploy gap V3 audited March 2023, launched April 2023 — approximately 30 days, within green threshold. Infinity: audits conducted 2024, launched April 28 2025. Exact audit completion date not determinable from public sources; if mid-2024, gap is 8-12 months, exceeding the 60-day green threshold and approaching 180-day red. Yellow applied due to ambiguity. RD-F-010 yellow Static-analyzer high-severity count No published Slither/Mythril/Semgrep output for PancakeSwap deployed contracts found in public sources. Static tool run not performed in this assessment. V3 Pool source review confirms no user-controlled delegatecall, no selfdestruct, and lock-modifier reentrancy on core functions. Five+ years of live V2/V3 operation without core-AMM exploit and 6 audit firms finding no critical unresolved issues provides medium-confidence proxy signal. Yellow assigned due to lack of published tool output. RD-F-015 yellow ERC-777/1155/721 hook without reentrancy guard Core AMM does not explicitly integrate ERC-777 tokensReceived or ERC-721 onReceived callbacks. However, the 2025 BCE/USDT ($679K) and OCA/USDC ($422K) pool drains both involved fee-on-transfer token interactions with the AMM — the V3 lock() partially mitigates reentrancy but doesn't prevent reserve accounting distortion via fee-on-transfer. This is a class risk inherent to permissionless AMM pools with arbitrary ERC-20 support. Yellow because the risk is present at the pool level even though no ERC-777 specific reentrancy was exploited. RD-F-183 yellow Bug bounty scope gap on highest-TVL contracts Immunefi Group 1 (highest reward tier) covers AMM V2, V3, StableSwap, MasterChef V2/V3, Smart Chef, Cake Pool. Infinity (V4) core contracts (Vault, CLPoolManager, BinPoolManager) — launched April 2025, $1.73B TVL — are in Group 2 ('other contracts not mentioned in group 1'), receiving lower reward priority. Not fully out of scope, but the newest and most architecturally complex surface is in a lower-priority bounty tier. Yellow rather than red because Group 2 still provides coverage. RD-F-020 gray EIP-712 domain separator missing chainId CAKE token ERC-20 permit uses EIP-712. Exact domain separator struct not inspected from BscScan — static read of contract source was not performed for the CAKE token's DOMAIN_SEPARATOR. PancakeSwap operates cross-chain with the same CAKE token, making chainId presence critical. Cannot grade without source inspection. Marked gray due to insufficient evidence. RD-F-021 n/a UUPS _authorizeUpgrade correctly permissioned V2 Factory, V3 Factory, V3 Pool are immutable non-upgradeable contracts with no proxy pattern. Infinity Vault and PoolManagers are singleton non-proxy deployments. MasterChef V2 is a direct deployment (Solidity 0.6.12). No UUPS proxy pattern detected across primary contracts. Factor N/A. RD-F-023 n/a Constructor calls _disableInitializers() No upgradeable proxy pattern confirmed for core PancakeSwap contracts. V2/V3/Infinity primary contracts use constructors with immutable deployments, not initializable proxy patterns. The factor's measurement (_disableInitializers() in constructor) applies only to proxied implementations. N/A for this protocol architecture.
RD-F-003 green Resolved-without-proof findings OtterSec publicly announced 2 high-impact Infinity vulnerabilities resolved prior to launch. The 2021 Lottery V2 vulnerability was disclosed via Immunefi and patched before exploitation (Immunefi post-mortem confirms). No post-mortem documents an auditor finding marked resolved that remained unpatched on-chain.
RD-F-004 green Audit count 6 distinct EVM audit firms: SlowMist, PeckShield, BlockSec, Hexens, OtterSec, Zellic. 4 Aptos audit firms: OtterSec, Zellic, Halborn, SlowMist. Infinity core alone audited by 3 firms (Hexens, OtterSec, Zellic). Clearly exceeds the 2-firm green threshold across all major surfaces.
RD-F-007 green Bug bounty presence & max payout Active Immunefi bug bounty with $1,000,000 maximum payout for critical smart contract vulnerabilities. Payout capped at 5% of funds at risk per report with $50,000 minimum for Group 1 critical. Group 1 covers AMM V2, V3, StableSwap, MasterChef V2/V3, Smart Chef, Cake Pool. Last updated 2026-03-16.
RD-F-008 green Ignored bounty disclosure The 2021 Lottery V2 vulnerability was disclosed via Immunefi and patched before exploitation — this was a successful bounty catch, not an ignored disclosure. The 2025 BCE/USDT exploit was a third-party token interaction issue (fee-on-transfer token design flaw), not a disclosure that was ignored by the team. No post-mortem documents an ignored pre-exploit disclosure on core AMM.
RD-F-011 green SELFDESTRUCT reachable from non-admin path V3 Pool source inspection confirms no SELFDESTRUCT opcode. V2 Factory/Pair (Solidity 0.5.16) confirmed no selfdestruct by 5+ years live operation without attack. Infinity contracts (0.8.26 Cancun) — selfdestruct is deprecated and non-functional in Cancun EVM. No auditor-reported selfdestruct finding across 6 firms.
RD-F-012 green delegatecall with user-controlled target V3 Pool source inspection: no user-controlled delegatecall. Callbacks (IPancakeV3MintCallback, IPancakeV3SwapCallback) are external calls to user-supplied addresses for flash repayment, not delegatecall — standard Uniswap V3 pattern. V2 Pair: no delegatecall. Infinity adapters use SafeCallback pattern. No auditor-reported user-controlled delegatecall across 6 firms.
RD-F-013 green Arbitrary call with user-controlled target V3 Pool callback pattern calls user-supplied swap/mint callee addresses for flash repayment verification — constrained by the lock() mutex ensuring invariants hold post-call. No unconstrained arbitrary call with user-supplied target and data detected. No auditor findings on this pattern across 6 firms covering the deployed code.
RD-F-014 green Reentrancy guard on external-calling functions V3 Pool uses lock() modifier (slot0.unlocked bool mutex) on mint, burn, swap, flash — all externally-calling state-mutating functions. V2 Pair uses lock modifier on swap, mint, burn. Infinity uses vault-lock mechanism. Multiple audits across 6 firms found no reentrancy issues in live functions.
RD-F-016 green Divide-before-multiply pattern V3 Pool uses FullMath library (fixed-point 256-bit arithmetic), TickMath, SqrtPriceMath derived from Uniswap V3's audited math libraries. No divide-before-multiply finding surfaced in 4 V3 audits (SlowMist x2, PeckShield x2). V2 uses k=xy invariant math with no complex division patterns.
RD-F-017 green Mixed-decimals math without explicit scaling V3 uses sqrtPriceX96 encoding which normalizes for token decimal differences. V2 uses reserve-based math with no cross-decimal arithmetic in core pair. StableSwap (Curve fork) is designed for same-decimal stable pairs. No auditor-reported decimal math issue across 6 firms.
RD-F-018 green Signed/unsigned arithmetic confusion V3 uses int256/uint256 casting with standard Uniswap V3 SafeCast patterns. V2 is pure uint arithmetic. No signed/unsigned confusion finding across 6 audit firms. Standard Uniswap-derived math is well-audited for this class of issue.
RD-F-019 green ecrecover zero-address return unchecked PancakeSwap V2/V3 AMM core contracts do not use ecrecover in core swap/liquidity paths. CAKE token uses ERC-20 Permit (EIP-2612). No auditor-reported ecrecover zero-address issue for core contracts across 6 firms. Deployed since 2020 with no reported ecrecover exploit.
RD-F-022 green Public initialize() without initializer modifier V2 PancakePair.initialize(address,address) is guarded by require(msg.sender == factory) — factory-only guard prevents unauthorized calls for this non-proxy architecture. V3 Pool has no initialize function (constructor-deployed). Infinity Vault has no proxy initialize pattern. MasterChef V2 uses a custom init() gated by onlyOwner. No exploitable unprotected initialize found on any live implementation.
RD-F-024 green Code complexity vs audit coverage V3 codebase is a Uniswap V3 fork — well-understood codebase with 4 dedicated audit engagements (SlowMist x2, PeckShield x2) in March-April 2023. Infinity has 3 firm audits (Hexens, OtterSec, Zellic) for both core and periphery components. No auditor noted insufficient time relative to code size in public disclosures.
Governance & admin Yellow 49 24 of 24
RD-F-036 red Flash-loanable voting weight Snapshot governance: 1 CAKE = 1 vote based on current wallet balance at snapshot block. No lock requirement, no checkpoint, no quorum required. veCAKE (time-weighted) retired May 2025 under Tokenomics 3.0, removing prior mitigation. Demonstrated in April 2025: 8 addresses locked ~25M CAKE (approx. 50% of unlocked supply) days before Tokenomics 3.0 vote — a direct voting manipulation event. No minimum quorum per docs. RD-F-037 red Quorum achievable via single-entity flash loan No quorum threshold exists per governance docs. Any token balance at snapshot block sufficient to pass a proposal. CAKE available in DEX liquidity pools (PancakeSwap V3 BSC). Large holder or coordinated entity can control vote outcome — demonstrated by April 2025 Tokenomics 3.0 manipulation event. RD-F-038 red Proposal execution delay < 24h Timelock delay is 21,700 seconds (~6 hours), well below the 24h threshold. Proposals can be queued and executed within 6 hours with no additional delay. RD-F-042 red Admin has mint() with unlimited max CAKE token mint() restricted to onlyOwner (MasterChef). MasterChef has no per-call mint cap at the contract level. Governance-level 400M hard cap (passed Jan 2026) is a social/governance constraint, not enforced in contract code. Contract-level mint authority is unlimited subject only to the governance cap. MasterChef owner is unconfirmed (likely team multisig) but holds effective CAKE inflation authority. RD-F-047 red Governance token concentration (Gini) Plutocratic 1-CAKE-1-vote governance with no time-weighting (veCAKE retired May 2025). April 2025 manipulation: 8 addresses controlled ~25M CAKE (~50% unlocked supply) pre-vote. No quorum. De facto concentration is high. Gini coefficient not formally computed but documented manipulation event demonstrates single-entity domination feasibility. RD-F-025 yellow Admin key custody type Hybrid custody: team-claimed multisig + legacy Timelock (6h delay). No on-chain Governor. Snapshot signaling with team execution. Admin for CAKE token = MasterChef; MasterChef owner = unconfirmed. FAQ states 'chefs use multisig for all contracts' but Safe API returned null and no Safe address identified. RD-F-027 yellow Single admin EOA Timelock initialized with admin EOA 0x11904A8e60Bd756f48d20a6970ff86B7A43bC022 (last active August 1 2021; 1,731+ days dormant). Current admin path for active contracts unconfirmed as multisig. Team FAQ claims multisig usage. Not confirmed red (single EOA) because team attestation exists; not confirmed green because no multisig address verified on-chain. RD-F-032 yellow Timelock duration on upgrades Timelock delay is 21,700 seconds (~6.03 hours). Below the 24-48h peer standard at $1.73B TVL. Aave uses 1-day, Uniswap 2-day, Compound 2-day timelocks. A 6-hour window is insufficient for community reaction to a malicious governance action at this TVL. RD-F-033 yellow Timelock on sensitive actions Timelock (6h) covers governance-directed changes. However: (1) 6h is thin at $1.73B TVL; (2) current timelock admin is a dormant EOA (last active Aug 2021); (3) Infinity/V4 contracts may have separate admin paths not going through this legacy timelock; (4) not all sensitive functions (mint, rescue) confirmed to route through timelock. RD-F-034 yellow Guardian/pause-keeper distinct from upgrader No formal guardian/pause-keeper role identified with a separate multisig or on-chain mechanism. Team retains emergency intervention rights per governance docs ('team reserves the right to intervene in critical situations without requiring a community vote') but this is an undifferentiated team power, not a distinct guardian role. RD-F-035 yellow Role separation: upgrade ≠ fee ≠ oracle Core V2/V3 AMM contracts are immutable — no upgrade role exists. Fee management and oracle config (for Predictions) are separate. However, at the governance layer, the team multisig appears to control all admin functions without documented role separation. Treasury, upgrade, and oracle config roles appear to converge in the undifferentiated team admin. RD-F-040 yellow Emergency-veto multisig present Team claims emergency intervention rights without community vote. No formal on-chain emergency-veto multisig identified. Governance docs: 'team reserves the right to intervene in critical situations — such as security threats or issues affecting stable operation — without requiring a community vote.' This is team discretion, not a formal on-chain veto mechanism. RD-F-041 yellow Rescue/emergencyWithdraw without timelock Timelock exists at 0xA1f482Dc (6h delay). MasterChef V2 (CAKE minting authority) uses a Timelock path for changes. However: 6h is extremely thin at $1.73B TVL; current Timelock admin is a dormant EOA (last active Aug 2021); exact coverage of all rescue/emergency paths through this timelock not confirmed. BCE/USDT pool was paused post-exploit manually, suggesting team can make immediate operational changes. RD-F-045 yellow Constructor args match governance proposal Timelock constructor args visible on BscScan (delay=21700, admin=0x11904A8e60). No material divergence identified for the Timelock. Infinity/V4 constructor args not verified against any governance proposal. Partial confidence. RD-F-167 yellow Deprecated contract paused but pause reversible by live admin veCAKE (0x5692DB8177a81A6c6afc8084c2976c9933ec1bab) deprecated May 2025 with redemption window until October 2025. MasterChef V1 is legacy but still operational. Whether admin retains reversible pause over veCAKE post-deprecation is unknown. Legacy surfaces with residual admin authority represent tail risk. RD-F-026 gray Upgrade multisig signer configuration (M/N) Safe API returned null. Data cache safe_multisigs: []. No confirmed multisig address for PancakeSwap on BSC. Signer configuration (M/N) cannot be assessed. RD-F-028 gray Low-threshold multisig vs TVL No confirmed multisig address. Safe API gap. Cannot determine threshold vs TVL peer cohort. Safe_multisigs: [] in data cache. RD-F-029 gray Multisig signers co-hosted Cannot assess — no confirmed multisig address to enumerate signers. RD-F-030 gray Hot-wallet signer flag Cannot assess — no confirmed multisig address. RD-F-031 gray Signer rotation recency Cannot assess signer rotation recency — no confirmed multisig address to check historical signer set changes. RD-F-039 n/a delegatecall/call in proposal execution without allowlist No on-chain Governor contract. Governance is Snapshot signaling (off-chain) with team multisig execution. There is no proposal executor contract capable of delegatecall. Factor N/A by construction. RD-F-044 gray Admin wallet interacts with flagged addresses Cannot assess — no confirmed admin address to check against watchlist.
RD-F-043 green Admin = deployer EOA after 7 days Protocol launched September 2020 (67+ months). Deployer EOA 0x0F9399FC81DAC77908A2DDE54BB87EE2D17A3373 is not current admin — ownership transferred historically. The 7-day window is moot given protocol age.
RD-F-046 green Contract unverified on Etherscan/Sourcify All primary BSC contracts verified on BscScan: CAKE token, Factory V2, Router V2, Factory V3, MasterChef V1/V2 all have verified source code. Infinity core and periphery hosted on GitHub with audit PDFs. No unverified contracts at launch.
Oracle & external dependencies Yellow 21 17 of 17
RD-F-049 yellow Oracle role per asset Chainlink is Primary for Predictions (BNB/BTC/ETH). Triple-oracle (Pyth primary, Binance Oracle secondary, Chainlink tertiary) for ApolloX perpetuals. AMM has no oracle role. Fallback for Predictions is round-cancellation + refund (not secondary feed). Exact priority-ordering of triple-oracle not verified at source level. RD-F-050 yellow Dependency graph (protocols depended upon) Dependencies mapped: (1) Chainlink 14 BSC feeds (Predictions/auxiliary); (2) ApolloX/APX Finance — entire perpetuals product; (3) LayerZero V1 oracle + relayer — CAKE bridge; (4) Chainlink VRF historical (Lottery V2). Key risk: ApolloX dependency means a security incident at ApolloX directly affects PancakeSwap perps users. Infinity/V4 hooks could add additional deps (not yet assessed). RD-F-051 yellow Fallback behavior on oracle failure Predictions: oracle failure → round cancelled, users refunded (no fund loss, UX disruption). Perpetuals (ApolloX): triple-oracle provides implicit fallback — documented that it 'dodges unexpected price behaviours.' AMM: no oracle, no fallback needed. Other Chainlink-consuming functions (AVAX/USD 86400s, COMP/USD 86400s): no documented fallback. RD-F-052 yellow Breakage analysis per dependency Per-dependency breakage: Chainlink Predictions feeds fail → rounds cancelled, refunds. ApolloX fails → perpetuals halt, PancakeSwap has no control. LayerZero V1 oracle/relayer fails → CAKE bridging halts. AVAX/USD 86400s heartbeat → 24h stale price possible. Analysis incomplete at ApolloX source level. RD-F-057 yellow Circuit breaker on price deviation No automatic price-deviation circuit-breaker documented for Predictions (manual pause only; oracle failure → round cancellation). Perpetuals: triple-oracle model provides implicit protection against single-source price manipulation, but programmatic circuit-breaker threshold not verified at ApolloX source level. RD-F-059 yellow Oracle staleness check present Predictions: oracleUpdateAllowance constructor param confirms staleness mechanism intent. Docs note oracle refreshes ~every 20s; rounds cancel if oracle not updated by round end. Implementation completeness not fully verified (source truncated on BscScan). Concern: AVAX/USD (86400s heartbeat) and COMP/USD (86400s) — 24h stale price window possible for feeds with these heartbeats. RD-F-060 yellow Chainlink aggregator min/max bound misconfig BSC Chainlink feed proxies confirmed as EACAggregatorProxy (standard Chainlink). Min/max bounds set in underlying aggregator (not the proxy). Bounds require direct aggregator state read. Note: some feed proxy addresses show creation ~103 days ago (approximately January 2026), which may indicate recently updated proxy addresses — normal for Chainlink but worth verifying these are canonical production addresses. RD-F-062 yellow External keeper/relayer not redundant LayerZero V1 default relayer (LayerZero Labs) is a single-relayer dependency for CAKE bridging. If LayerZero Labs' relayer goes down, CAKE cross-chain transfers freeze. Predictions: Chainlink oracle network is multi-node (not a single keeper), but prediction round execution may depend on a keeper calling executeRound(). ApolloX manages its own keeper for perps. Rate-limiting caps exist for bridge (partial mitigant). RD-F-054 n/a TWAP window duration PancakeSwap's oracle-consuming functions use push-oracles (Chainlink, Pyth, Binance Oracle), not TWAP-based oracles. AMM V3 concentrated liquidity uses tick-math for pricing, not a TWAP read for its own price calculation. TWAP window duration factor is N/A. RD-F-055 n/a Oracle pool depth (USD) No DEX-TWAP oracle used by PancakeSwap's oracle-consuming functions. Oracle pool depth is N/A. RD-F-058 n/a Max-deviation threshold (bps) No max-deviation threshold (bps) parameter confirmed at source level for Predictions contract. PredictionV2 BscScan shows oracleUpdateAllowance (staleness) but no maxDeviationBps field visible from partial source. Perpetuals (ApolloX): not documented. Cannot assess without complete source code read. RD-F-181 n/a Permissionless-pool lending oracle PancakeSwap is a DEX, not a lending protocol. No borrow/collateral mechanics in V2/V3/StableSwap core. Data cache: borrow.present=false. F181 (permissionless-pool lending oracle) is N/A for non-lending DEX protocols per datacollectionguide §6 pre-mark.
RD-F-048 green Oracle providers used 14 Chainlink BSC feeds confirmed in data cache (BTC/USD, ETH/USD, BNB/USD, USDC/USD, USDT/USD, LINK/USD, AVAX/USD, COMP/USD, UNI/USD). Pyth + Binance Oracle + Chainlink triple-oracle for ApolloX perpetuals confirmed via docs. AMM core uses no external oracle.
RD-F-053 green Oracle source = spot DEX pool (no TWAP) N/A for AMM swap routing (canonical EVM DEX pre-mark — no oracle used for pricing in AMM). Predictions: Chainlink push-oracle (not spot DEX). Perpetuals: triple push-oracle (Pyth + Binance Oracle + Chainlink). No raw spot DEX price drives any oracle-gated function.
RD-F-056 green Single-pool oracle (no medianization) Perpetuals uses triple-oracle (Pyth + Binance Oracle + Chainlink) — not a single-pool oracle. Predictions uses Chainlink as primary (single source, but Chainlink is multi-node aggregated, not a single DEX pool). AMM: no oracle. Single-pool oracle pattern does not apply.
RD-F-061 green LP token balanceOf used for pricing AMM V2/V3 pricing is based on the reserve ratio from pool contract internal state. Standard Uniswap V2/V3 AMM design — does not use balanceOf of LP tokens in a separate contract for pricing. Donation to a V2 pool changes reserves and therefore price, but this is the standard AMM mechanism, not the LP-token balanceOf pricing pattern this factor targets.
RD-F-180 green Immutable oracle address [★ F180 — PD-017 T-14 tracking] PredictionV2 oracle stored as 'AggregatorV3Interface public oracle' — mutable public state, NOT immutable keyword. Constructor accepts _oracleAddress parameter. AMM core has no oracle address. ApolloX perpetuals uses external protocol oracle management. No immutable oracle address pattern found. Admin replaceability confirmed from non-immutable public storage.
Economic risk Yellow 25 13 of 13
RD-F-064 yellow TVL concentration (top-10 wallet share) 99.67% of TVL on BSC (BNB Chain) — a single-chain concentration that would cascade near-total protocol TVL in any BSC-level systemic event. Top-10 LP wallet share not precisely quantifiable (on-chain scan not available via WebFetch; Dune returns 403). Governance vote concentration documented: 8 addresses locked ~25M CAKE before Tokenomics 3.0 vote (April 2025), representing approximately half of then-unlocked supply — demonstrates that governance token concentration translates to economic parameter control. RD-F-065 yellow Liquidity depth per major asset Major pairs (BNB/USDT, BNB/USDC, CAKE/BNB) have deep liquidity — $50K BNB/USDT swap at <0.05% price impact; PancakeSwap is the dominant on-chain venue on BSC. However, permissionless pool creation creates a persistent class of thin long-tail pools. BCE/USDT pool (~$679K drained, March 2025) and OCA/USDC pool (~$422K drained) demonstrate that thin-pool long-tail exposure is a recurring harm surface. Net yellow because the tail risk from thin pools is material and recurrent, even if major pairs are deep. RD-F-072 yellow Market-listing governance threshold V2/V3 pool creation is fully permissionless — any BEP-20 token can be listed by adding liquidity; no governance approval, no token vetting, no minimum liquidity floor enforced by protocol code. This is the structural root cause of the BCE/USDT ($679K, March 2025) and OCA/USDC ($422K) pool drain incidents, where malicious/flawed third-party tokens were permissionlessly listed and exploited. PancakeSwap's default token list provides UI-level curation only; the on-chain contracts impose zero listing controls. Note: PD-024 marks RD-F-072 as 'lending-only N/A for DEX' — however, the economic harm pattern (permissionless market creation enabling exploitation) is structurally identical. Scored yellow rather than N/A given two documented dollar-denominated events. Flag as potential v1.6 scope clarification. RD-F-066 n/a Utilization rate (lending protocols) PancakeSwap is a DEX (AMM), not a lending protocol. No borrow/supply accounting. Data cache confirms borrow.present: false. N/A per PD-024 resolution (lending-only factor). RD-F-067 n/a Historical bad-debt events Not applicable in the lending sense — PancakeSwap does not accumulate bad debt. BCE/USDT ($679K, March 2025) and OCA/USDC ($422K) pool drains are LP impermanent-loss-style events from third-party token mechanics, not protocol bad debt events. The core AMM invariant was not broken. N/A per PD-024. RD-F-068 n/a Collateralization under stress PancakeSwap is a DEX, not a CDP or lending protocol. No collateral positions. N/A per PD-024. RD-F-069 n/a Algorithmic / under-collateralized stablecoin PancakeSwap does not issue stablecoins. CAKE is a governance/utility token with a deflationary emission model (>102% of minted CAKE burned weekly as of July 2025). No algorithmic stable design. N/A per PD-024. RD-F-070 n/a Empty cToken-style market (zero supply/borrow) [★] Not applicable. PancakeSwap is a Uniswap V2/V3/StableSwap fork DEX — not a Compound V2 fork lending protocol. No cToken-style markets. No totalSupply/totalBorrow accounting. The BCE/USDT March 2025 exploit was a burn-mechanism manipulation via a third-party token, structurally distinct from an empty-market donation attack. Profile §5 confirms fork lineage is Uniswap V2, Uniswap V3, Curve, and Uniswap V4 — no Compound fork lineage exists. Data cache lending_protocol: false. RD-F-071 n/a Seed-deposit requirement for new market listing Not applicable — not a lending protocol. No lending-market seed deposit concept. N/A per PD-024. RD-F-073 n/a Oracle-manipulation-proof borrow cap Not applicable — PancakeSwap is a DEX with no borrow caps. AMM V2/V3 core pricing is derived from on-chain pool reserves, not external oracles. Chainlink feeds in data cache serve Predictions/Perpetuals products. N/A per PD-024. RD-F-074 n/a ERC-4626 virtual-share offset (OZ ≥4.9) PancakeSwap does not use ERC-4626 vaults in its core AMM. V2/V3 LP tokens use the Uniswap LP token model (ERC-20 with pool-share accounting). Infinity/V4 singleton Vault uses flash accounting, not ERC-4626 share mechanics. N/A per PD-024. RD-F-075 n/a First-depositor / share-inflation guard Not applicable as defined for lending/ERC-4626 vaults. AMM V2/V3 x*y=k invariant is resistant to first-depositor inflation in the lending sense — no cToken minting. Thin-pool risk from minimal initial liquidity is a related but distinct risk captured under F064/F065. N/A per PD-024.
RD-F-063 green TVL (current + 30d trend) $1.73B TVL as of 2026-04-28 (data cache); live refetch shows ~$2.075B. 30-day trend +2.78%. 12-month peak ~$2.6B (Sep 2024). Exceeds $100M floor by 17x. Positive recovery trend.
Operational history Yellow 29 15 of 15
RD-F-082 red Post-mortem published within 30 days No PancakeSwap-authored post-mortem published for BCE/USDT March 2025 incident as of 2026-04-28 (13+ months post-incident). Third-party analyses exist (BlockSec, Medium independent post). The 2021 Lottery Immunefi post-mortem was for a bounty disclosure, not an exploit. Threshold: red = >90 days or not published. RD-F-089 red Insurance coverage active No active Nexus Mutual, Sherlock, or Unslashed coverage confirmed for PancakeSwap AMM contracts. Immunefi bug bounty ($1M cap) is not insurance coverage. $1M cap on $1.73B TVL = 0.058% ratio. No active cover panel found via Nexus Mutual or Sherlock searches. Near-universal for large DEX protocols. RD-F-077 yellow Prior exploit count 1 confirmed on-chain pool incident: BCE/USDT V3 pool exploit March 2025 (~$679K loss; root cause is BCE token's malicious fee-on-transfer burn-to-pair logic, not AMM code). No recovery confirmed. 2021 Lottery vulnerability was a successful bug-bounty disclosure (not an exploitation). 2021 DNS hijack and 2025 X hack are off-chain incidents not counted. Conservative yellow: 1 incident, no recovery. RD-F-081 yellow Post-exploit response score BCE/USDT 2025 response: pool pause executed same day (fast). No user compensation program. No PancakeSwap-authored post-mortem. Root cause identified only by third-party researchers (BlockSec, independent Medium analysis). Operational recovery: broader protocol remained live; only affected pool paused. Score: 2/5 — fast operational response, poor transparency/compensation. Yellow threshold: 2–3/5. RD-F-084 yellow TVL stability (CoV over 90d) DefiLlama API trailing 90-day TVL: mean ~$1,760M, std dev ~$310M, CoV ≈ 0.176 (17.6%). Threshold: green <0.15; yellow 0.15–0.35. CoV 0.18 falls in yellow band. Range driven by broad BSC market beta, not protocol-specific operational issues. Assessment date: 2026-04-28. RD-F-085 yellow Incident response time (minutes) BCE/USDT 2025: PancakeSwap team paused affected pools same day as exploit (2025-03-15); BlockSec confirmed same day. Exact response time in minutes not determinable from public sources (no pause tx hash/block timestamp retrieved without RPC). Estimated hours (same-day, likely <4h). Yellow: 60 min to 4 hours estimated. Cannot confirm green (<60 min) without on-chain tx timestamp. RD-F-086 yellow Pause activations (trailing 12 months) BCE/USDT pool was temporarily paused in March 2025 (within trailing 12 months ending 2026-04-28). 1 documented pause activation with documented reason (exploit response). No other pause activations identified. Threshold: yellow = 1–2 pauses with documented reason. RD-F-166 yellow Deprecated contracts still holding value MasterChef V1 (0x73feaa1eE314F8c655E354234017bE2193C9E24E, BSC) superseded by MasterChef V2 April 2022. As of 2026-04-27: holds 7,332,676 CAKE (~$10.9M at ~$1.49/CAKE) + 113 additional BEP-20 tokens. Last transaction 2026-04-27 (users still actively staking/unstaking). $10.9M well above $100K materiality threshold. Team announced migration but did not force-drain; voluntary exit model. Yellow: deprecated by intent, still holds material value. RD-F-083 gray Auditor re-engaged after last exploit The BCE/USDT 2025 incident root cause is a malicious third-party token's own logic (BCE scheduledDestruction), not a bug in PancakeSwap AMM code. A protocol re-audit specifically addressing this root cause is not applicable — the AMM invariant math was not implicated. No post-incident external audit for the AMM specifically commissioned. Score: gray (N/A — incident did not implicate protocol-owned smart contract code in a way requiring a re-audit).
RD-F-076 green Protocol age (days) First mainnet deploy 2020-09-20 (BSC). As of 2026-04-28: 2,047 days live (~67 months). Well above the green threshold of ≥365 days.
RD-F-078 green Chronic-exploit flag (≥3 incidents) 1 confirmed on-chain exploit (BCE/USDT 2025). Off-chain incidents (DNS hijack 2021, X hack 2025) do not count. Total = 1, well below ≥3 CHRONIC threshold. CHRONIC flag does not fire.
RD-F-079 green Same-root-cause repeat exploit Only 1 confirmed on-chain pool incident (BCE/USDT 2025; root cause: malicious fee-on-transfer token). No second incident with same root-cause cluster. DNS hijack (2021) and X hack (2025) share a social-engineering/account-compromise class but are off-chain and caused no on-chain fund loss. No same-root-cause repeat exploit.
RD-F-080 green Days since last exploit Most recent on-chain pool incident: 2025-03-15 (BCE/USDT). Days since: 409 days as of 2026-04-28. Green threshold: >365 days. Just clears green.
RD-F-087 green Pause > 7 consecutive days The BCE/USDT pool pause was a targeted pool-level action (not a full-protocol pause). No evidence of any protocol-wide pause exceeding 7 consecutive days in the trailing 12 months. Protocol (other pools, V2/V3/StableSwap) remained fully operational throughout the March 2025 incident period.
RD-F-088 green Re-deployed to new addresses in last year Infinity (V4) launched 2025-04-28 as an additive new product — not a forced migration retiring existing contract addresses. No evidence of V2/V3 address retirement in trailing 12 months. MasterChef V1 deprecation is historical (2022). No redeploy-to-new-addresses event in last 12 months.
Real-time signals Green 17 22 of 22
RD-F-100 yellow Flash loan >$10M targeting protocol tokens PancakeSwap V3 provides flash loan functionality natively. Protocol is routinely used as flash-loan liquidity source in BSC-ecosystem attacks against OTHER protocols (Dec 2025 MSCST exploit used PancakeSwap pools as oracle manipulation venue). Core PancakeSwap AMM has not been subject to a protocol-level flash-loan attack. The risk surface is structurally high: $1.73B TVL, permissionless pools, used as primary DEX in DPRK laundering flows. Yellow reflects elevated structural exposure as a flash-loan infrastructure node even though no current protocol-level attack is detected. Phase-2 signal. RD-F-103 yellow Bridge signer-set change proposed/executed CAKE LayerZero OFT bridge confirmed via external sources (BSC ProxyOFT 0xb274202daBA6AE180c665B4fbE59857b7c3a8091). Data cache shows layerzero.present: false — confirmed pipeline gap. Cannot verify current LayerZero DVN signer configuration or whether an unscheduled signer-set change has occurred. Signal applicable (Tier-A) but monitoring infrastructure gap means the signal is not yet wired. Yellow reflects unverified bridge configuration status. RD-F-105 yellow DNS/CDN/frontend hash drift Elevated historical risk posture. 2021 GoDaddy DNS hijack confirmed (CoinDesk, Mar 2021) — direct compromise of pancakeswap.finance DNS through GoDaddy account takeover. Domain has since migrated to MarkMonitor (enterprise-grade registrar, Fortune 500 standard, substantially higher security than GoDaddy consumer tier). No current DNS drift detected. Active typosquat ecosystem: web2.pancake.run is a live drainer portal flagged by 14 security vendors as of 2026. Monitoring infrastructure (phase-2 signal) not yet deployed. Domain registrar improvement is material mitigation, but ongoing typosquat activity and historical precedent warrant yellow. RD-F-090 gray Mixer withdrawal → protocol interaction v1-phase-2 signal (Tier-C). Historically confirmed: Lazarus Group wallets laundered $263M of Bybit hack proceeds through PancakeSwap (Feb-Mar 2025) — past the 30-day lookback threshold as of 2026-04-28. As a permissionless BSC DEX, ongoing baseline mixer-to-protocol flows are structurally inherent but no specific live fire confirmed via public proxy sources today. Historical interaction is captured under RD-F-158 (Cat 11). RD-F-091 n/a Partial-drain test transactions v1-deferred per T-09 taxonomy. PH signal requiring specialized mempool tooling and pattern-matching infrastructure. Not in v1 production scope. RD-F-092 n/a Unusual mempool pattern from deployer wallet v1-deferred per T-09. P2 mempool signal requiring deployer-wallet behavior baseline. Deployer wallet 0x0F9399FC81DAC77908A2DDE54BB87EE2D17A3373 appears inactive in recent periods. RD-F-093 n/a Abnormal gas-price willingness from attacker wallet v1-deferred. P2 mempool gas-price signal. Not in production scope. RD-F-094 n/a New contract with similar bytecode to exploit template v1-deferred. P2 bytecode-similarity deploy scan signal. Not in production scope. RD-F-095 n/a Known-exploit function-selector replay v1-deferred. P2 function-selector replay pattern signal. Not in production scope. RD-F-096 n/a New ERC-20 approval to unverified contract from whale v1-deferred. P2 user-level approval signal, not protocol-level. Not in production scope. RD-F-097 n/a Sybil surge of identical-pattern transactions v1-deferred. P2 sybil-surge detection signal. Not in production scope. RD-F-101 gray Large governance proposal queued Signal partially inapplicable. No on-chain Governor contract detected (governor_address: null in data cache). Governance is Snapshot-only (cakevote.eth) — off-chain signaling only. RD-F-101 triggers on ProposalCreated events from on-chain Governor contracts. No such contracts exist for PancakeSwap. Admin execution is via team multisig (covered by RD-F-102). The April 2025 whale CAKE acquisition for voting power was unusual but does not trigger the flagged-pattern criteria (no calldata containing upgrade/admin-change selectors; no on-chain proposal with flash-loanable weight path). RD-F-102 gray Admin/upgrade transaction in mempool Phase-2 signal (mempool listener not deployed). Admin Timelock at 0xA1f482Dc58145Ba2210bC21878Ca34000E2e8fE4 (BSC, ~6-hour delay). Legacy admin EOA 0x11904A8e60Bd756f48d20a6970ff86B7A43bC022 is dormant since August 1, 2021 (1,731+ days inactive). Current admin multisig identity unconfirmed — cannot establish admin allowlist for mempool monitoring. No public reports of suspicious admin transactions detected. RD-F-106 n/a Cross-chain bridge unverified mint pattern v1-deferred. P1 cross-chain bridge tx pattern signal requiring cross-chain indexing infrastructure. PancakeSwap has bridge surface (CAKE LayerZero OFT) but this signal is not in v1 production scope. RD-F-107 n/a Admin EOA signing from new geography/device v1-deferred. P2 admin EOA geo/device fingerprint signal. Off-chain signing telemetry not publicly available. Not in v1 production scope. RD-F-108 n/a GitHub force-push to sensitive branch v1-deferred. P2 GitHub force-push monitoring signal. GitHub repo pancake-v3-contracts last commit 2024-04-22. No force-push events detected in public sources. Monitoring infrastructure not deployed. RD-F-109 n/a Social-media impersonation scam spike v1-deferred signal. However, this signal has HISTORICALLY FIRED for PancakeSwap: October 2025 Chinese X account compromise — attackers promoted fake 'Sir Pancake' token, directing users to wallet-draining phishing sites, generating $20M+ in fake trading volume. Active typosquat ecosystem (web2.pancake.run live drainer confirmed 2026). PancakeSwap is a persistent high-value target given $1.73B TVL and large retail BSC user base. Elevated posture acknowledged even though v1 signal monitoring is not deployed. RD-F-110 n/a Unusual pending/executed proposal ratio v1-deferred. P2 governance proposal ratio signal. Derived from RD-F-101 which is partially inapplicable (no on-chain Governor). Snapshot governance proposals are off-chain and cannot be monitored for unusual proposal ratios via on-chain event scanning. RD-F-182 gray Security-Council threshold reduction (RT) Batch-24 signal (T-09 v1.1 candidate pending FP-rate review). PancakeSwap does not have a formal Security Council structure analogous to Drift Protocol. The nearest analog is the admin Timelock (~6-hour delay, below peer norm) and team multisig. No threshold reduction event detected in public sources. The pre-existing ~6-hour Timelock delay is a static Cat 2 concern (RD-F-032), not a triggered RD-F-182 event. Current multisig signer composition unconfirmed — monitoring infrastructure not deployable without confirmed admin addresses.
RD-F-098 green TVL anomaly — % drop in <1h TVL $1,733,496,752 as of 2026-04-28T18:35:44Z. 1d change: -0.01%; 30d change: +2.78%. No drain detected. BSC 99.67% ($1,727,786,947). Threshold: TVL_now/TVL_baseline_30d < 0.70 within 60-min window — not met. Signal applicable and would NOT fire today.
RD-F-099 green Oracle price deviation >X% from secondary AMM core (V2/V3) uses reserves-based pricing — no external oracle for core swap operations. Chainlink feeds detected are for Predictions/Perpetuals products, not safety-critical AMM reads. No oracle deviation detected on the 14 enumerated BSC Chainlink feeds. Signal partially inapplicable to core AMM; applicable to Predictions/Perpetuals peripheral products. Phase-2 signal (monitoring infrastructure not deployed). No fires detected.
RD-F-104 green Stablecoin depeg >2% on shared-LP venue No active stablecoin depeg above 2% threshold as of 2026-04-28. USDT dominates BSC stablecoin supply (~60%). Chainlink USDT/USD feed (0x501e21126486424567f40D490856094D72986E41, heartbeat 900s, deviation 0.1%) and USDC/USD feed (0x51597f405303C4377E36123cBc172b13269EA163) showing normal operation per data cache. Signal applicable (USDT/USDC pair TVL is substantial). Would NOT fire today.
Dev identity & insider risk Green 18 16 of 16
RD-F-117 red ENS/NameStone identity bound to deployer Deployer address 0x0F9399FC81DaC77908A2Dde54Bb87Ee2D17a3373 has no ENS or NameStone name bound to it. BscScan label 'Contract Deployer' is a service label, not an ENS resolution. Funder address carries Space ID domain 'fiowind.bnb' but that is the funder, not the deployer. No identity-binding mechanism on the deployer address. RD-F-111 yellow Team doxx status Team is entirely pseudonymous with consistent chef-themed aliases (Hops, Thumper, ChefSnoopy, ChefMist) since September 2020. No real-name doxxing of any team member found. Category: consistent-pseudonym-with-track-record (5.5-year BSC operational history with Binance Labs backing). Factor is yellow because accountability depth is thin at $1.73B TVL despite long track record. RD-F-112 yellow Team public accountability surface Public accountability surface is very thin. @HopsPancakeSwap X account exists (active since ~2021) but no real-name, employer, or location disclosed. GitHub contributors (ChefSnoopy, ChefMist, chef-omelette) are pseudonymous with no LinkedIn, conference talks, or prior-employer disclosures identified. Count of verifiable public trails per team member: 0-1. Yellow reflects inadequate accountability surface at $1.73B TVL scale. RD-F-121 yellow Contributor OSINT depth score Curator OSINT depth score for key identifiable team members: @HopsPancakeSwap — X account only, no LinkedIn, no conference presence (score: 1/5). ChefSnoopy/ChefMist — GitHub accounts only (score: 1/5). No team member scores above 1/5. Average score: 1/5. Yellow per rubric (average 2-3 = yellow; average 1 = red). Scored yellow rather than red because the protocol has 5.5 years of operational presence which itself functions as a weak public trail, and the pseudonymity is by design not by evasion. RD-F-123 yellow Sudden admin-rescue/ACL change without discussion Tokenomics 3.0 (April/May 2025): Forum thread opened April 8, 2025 with 197+ community replies; Snapshot vote on cakevote.eth preceded on-chain execution. This major ACL change (veCAKE sunset, emissions control to Kitchen) had substantial public discussion. No sudden admin-rescue events identified in last 180 days without discussion. Residual yellow: Timelock admin EOA (0x11904A8e60Bd756f48d20a6970ff86B7A43bC022) last active 2021; current admin for Infinity/V4 contracts unconfirmed; unable to verify whether all admin transitions had corresponding discussions. RD-F-116 gray Contributor tenure at admin-permissioned PR Cannot determine contributor tenure at time of admin-permissioned commits. GitHub org has zero public members; org membership join dates not accessible. ChefSnoopy (most recent commits: Apr 2024) and ChefMist (Feb 2024) commit history visible but account-creation dates and org-join dates not determinable without API access to private org data. Factor is not_assessed due to data unavailability. RD-F-119 gray Commit timezone consistent with stated geography GitHub commit timestamps display dates only, no UTC offset visible. Commit pattern is sparse (pancake-v3-contracts: Apr 2024, Feb 2024, Apr 2023; pancake-smart-contracts: Mar 2024, Feb 2024, Apr 2022). Cannot perform timezone analysis without time-of-day data. No DPRK UTC+9 anomaly identified. Team geography not publicly stated. RD-F-120 gray Video-off/voice-consistency flag No public video interviews, AMA recordings, or voice-validated public appearances found for any named PancakeSwap team member. Team operates entirely via text channels (X, blog, governance forum). Cannot assess video-off/voice-consistency. Gray is appropriate.
RD-F-113 green Team other-protocol involvement history No prior protocol roles identified for any named team member. GitHub contributors show no other major protocol deployments. No prior rug or exit-scam associations found in OSINT or rekt.news. data cache confirms rekt.incidents: []. 5.5-year clean operational history is the primary positive signal.
RD-F-114 green Deployer address prior on-chain history Deployer 0x0F9399FC81DaC77908A2Dde54Bb87Ee2D17a3373: ~1,025 BscScan transactions; first tx ~October 2019 (pre-BSC launch); latest tx ~March 2022. BscScan label: 'Contract Deployer'. No linked-to-prior-rug history. Funded by fiowind.bnb, a clean retail DeFi wallet. Category: normal-dev-history.
RD-F-115 green Prior rug/exit-scam affiliation No team member linked to a prior rug or exit-scam labeled protocol found across all OSINT sources. Unverified 2021 Medium article alleging Syrup exploit cover-up is anonymous opinion without tx-hash evidence — not credited. No Rekt.news entries for PancakeSwap team members. 5.5-year no-rug operational track record.
RD-F-118 green Handle reuse across failed/rugged projects @HopsPancakeSwap X handle: active under consistent PancakeSwap branding since at least 2021, no evidence of prior association with failed/rugged projects. Chef-class GitHub handles not found associated with prior rugged protocols. No handle-reuse pattern identified.
RD-F-122 green Contributor paid to DPRK-cluster wallet No evidence of contributor payments routing to DPRK-labeled wallets. The Allium/Warren finding (Dec 2025 Senator letter) is about Lazarus Group using PancakeSwap platform for money laundering, not about developer payment flows to DPRK cluster. ZachXBT DPRK IT worker investigations do not name PancakeSwap as an employer of DPRK developers. No blockchain analytics report identifies any PancakeSwap contributor payment address linking to DPRK cluster.
RD-F-124 green Deployer wallet mixer-funded within 30 days [★ CRITICAL — GREEN] Deployer 0x0F9399FC81DaC77908A2Dde54Bb87Ee2D17a3373 funded by fiowind.bnb (0x83b7f4547401141f4c1fd21e86e3f72579bbe3ec). BscScan on funder: personal retail DeFi wallet with Space ID domain, clean labels, no Tornado Cash interaction, no mixer flags. Funding predates BSC mainnet launch (~Oct 2019 first tx). No mixer interaction detected for deployer in any 30-day window around deploy (September 2020). No Tornado Cash or equivalent mixer involvement.
RD-F-125 green Deployer linked within 3 hops to DPRK/Lazarus [★ CRITICAL — GREEN] Deployer 0x0F9399FC81DaC77908A2Dde54Bb87Ee2D17a3373: hop-1 funder (fiowind.bnb) is a clean retail DeFi wallet with no DPRK labels. Senator Warren Dec 2025 letter (Allium research) establishes Lazarus laundered ~$263M via PancakeSwap pools from Bybit hack — this is platform-use laundering, NOT a hop-path from the deployer to the Lazarus cluster. RD-F-125 definition requires on-chain path from deployer ≤3 hops to DPRK cluster; Lazarus transacting through PancakeSwap DEX pools does not create such a path. No DPRK proximity confirmed. No DPRK escalation triggered. Confidence medium (not high) — no Chainalysis/TRM feed available for full hop analysis.
RD-F-184 green Real-capital social-engineering persona No evidence of a 'team contributor' or 'external integrator' persona deploying ≥$1M real capital to PancakeSwap to build credibility ahead of social-engineering attack. The Oct 2025 X account compromise was an ATM-style phishing attack on the Chinese-language X account — not a long-running real-capital social-engineering infiltration (Drift/UNC4736 pattern). Lazarus use of PancakeSwap pools for laundering is adversarial DEX use, not insider credibility building. No parallel to the Drift Protocol Apr 2026 UNC4736 infiltration pattern identified.
Fork / dependency lineage Green 7 10 of 10
RD-F-129 yellow Code divergence from upstream (%) V2: thin fork — CAKE fee token, BEP-20, 0.25% fee vs 0.30%; core AMM invariant logic identical; estimated <20% divergence. V3: clean copy post-BUSL expiry with fee tier differences; estimated <15% divergence from Uniswap V3. StableSwap: Solidity rewrite from Vyper Curve — language change implies high divergence at source level but semantically similar. Infinity: documented differences from Uniswap V4 (BNB native support, different hook interface) — estimated >20% divergence. Precise git-diff percentage not achievable from static read. RD-F-133 yellow Dependency manifest uses unpinned versions V3 periphery package.json: @openzeppelin/contracts 3.4.2-solc-0.7 (exact version, pinned — green for that dep). pancake-smart-contracts V2 uses @openzeppelin/test-helpers ^0.5.11 (unpinned, but test-only dep). V2 production contracts do not use OZ in core pair/factory. Infinity uses Foundry; library pinning via .gitmodules git-submodule SHAs not inspected — Foundry convention typically pins commits but was not verified. Yellow due to incomplete gitmodules inspection.
RD-F-126 green Is-a-fork-of Multi-fork lineage fully documented: V2 = Uniswap V2 (README explicit); V3 = Uniswap V3 (launched day-after BUSL expiry, multiple sources confirm); StableSwap = Curve-style (SlowMist Aug 2022 audit scope); Infinity = Uniswap V4-inspired (singleton vault, hooks — PancakeSwap developer docs confirm both similarities and differences); Aptos DEX = original Move codebase (not a fork).
RD-F-127 green Upstream patch not merged No public Uniswap V2 security patches were issued post-2020 — Uniswap V2 is immutable (no upgrade mechanism; patches would require a new deploy). No Uniswap V3 critical vulnerability patches issued by Uniswap Labs since the V3 launch. For Infinity vs Uniswap V4: PancakeSwap explicitly lists key differences (different hook interface, BNB native support) — custom enough that upstream V4 patches would not automatically apply. No pending upstream patch identified affecting PancakeSwap's deployed code.
RD-F-128 green Upstream vulnerability disclosure (last 90d) No public Uniswap V2 or V3 vulnerability disclosures in last 90 days (as of 2026-04-29). No Curve vulnerability disclosures affecting Solidity StableSwap forks — the July 2023 Curve Vyper bug does NOT apply to PancakeSwap StableSwap which is Solidity. No LayerZero OFT-specific security disclosures affecting the CAKE bridge detected in last 90 days.
RD-F-130 green Fork depth (generations from original audit) V2: depth 1 (direct fork of Uniswap V2, itself originally audited). V3: depth 1 (direct fork of Uniswap V3). StableSwap: depth 1 from Curve. Infinity: depth 1 from Uniswap V4 concept (custom implementation). No fork-of-fork-of-fork detected. All surfaces are direct first-generation forks.
RD-F-131 green Fork retains upstream audit coverage All major version surfaces have independent fresh audits of the PancakeSwap fork: V2 (SlowMist May 2021 of PancakeSwap-specific V2); V3 (SlowMist x2 + PeckShield x2 March-April 2023, explicitly covering the PancakeSwap V3 fork); StableSwap (SlowMist Aug 2022); Infinity (Hexens + OtterSec + Zellic 2024 of infinity-core and infinity-periphery). No surface relies solely on upstream audit.
RD-F-132 green Fork has different economic parameters than upstream V2: 0.25% fee vs Uniswap V2's 0.30% — parameter change covered in SlowMist May 2021 audit scope. V3: fee tiers (0.01%, 0.05%, 0.25%, 1%) vs Uniswap V3 standard — PeckShield/SlowMist March 2023 audits explicitly covered V3 PancakeSwap-specific configuration. No major parameter deviation creating un-audited economic risk identified.
RD-F-134 green Dependency had malicious-release incident (last 90d) No GHSA or npm security advisory found for key dependencies (hardhat, ethers v5, @openzeppelin/contracts 3.4.2-solc-0.7, @pancakeswap/common) in trailing 90 days. No PancakeSwap-specific malicious dependency advisory identified.
RD-F-135 green Shared-library version with known-vuln status V3: @openzeppelin/contracts 3.4.2-solc-0.7 — no active high/critical GHSA advisory for OZ 3.4.x affecting AMM patterns (no upgradeable proxy, no governance in V3 core). V2: no OZ dependency in core pair/factory. Infinity: Foundry-based with solc 0.8.26 — newer stack. No active library advisory identified for the deployed versions.
Post-deploy hygiene & change mgmt Green 18 13 of 13
RD-F-136 yellow Deployed bytecode matches signed release tag Primary contracts (V2/V3) are immutable and verified on BscScan. GitHub last commit date for pancake-v3-contracts: 2024-04-22. Exact commit-to-deployed-bytecode matching not performed. Infinity contracts deployed April 2025 — commit SHA to bytecode match not verified. RD-F-137 yellow Upgrade frequency (per 90 days) Primary AMM contracts are immutable (no upgrades). Peripheral contracts may have upgrades but not enumerated. Infinity/V4 newly deployed April 2025. No systematic upgrade frequency data available. Timelock shows last tx April 2021 — no recent governance-directed upgrades through the timelock. RD-F-139 yellow Post-audit code changes without re-audit Infinity/V4 launched April 28 2025 with 3 external auditors (Hexens, OtterSec, Zellic for both core and periphery — confirmed via GitHub). CertiK's involvement was a hooks security blog post, not a full engagement audit. Tokenomics 3.0 contract changes (May 2025) — veCAKE sunset, new staking — unconfirmed for re-audit coverage. Post-Tokenomics-3.0 CAKE supply cap changes (Jan 2026) are governance-level. Multi-firm Infinity audits are a meaningful mitigant. Yellow, not red. RD-F-145 yellow Deployed bytecode reproducibility Verified source code on BscScan for primary contracts. Reproducibility not formally tested against GitHub source. Infinity contracts in GitHub repos (pancakeswap/infinity-core, pancakeswap/infinity-periphery) with audit reports but build reproducibility unconfirmed. RD-F-168 yellow Stale-approval exposure on deprecated router veCAKE deprecated May 2025 with redemption window until October 2025. MasterChef V1 legacy but still operational. Router V2 still widely used. Significant historical user approvals to legacy V1 and V2 contracts expected given 5+ year history and large user base. Specific approval count not enumerated but structural yellow expected. RD-F-185 yellow Bridge rate-limiter / chain-pause as positive mitigant No automated rate-limiter on core AMM. BCE/USDT pool was manually paused by developers post-exploit (not automated). LayerZero OFT CAKE bridge does not have a documented per-window rate-limiter. BSC chain-level pause exists via validator set but is not a protocol-specific mechanism. Manual pause capability exists but is not a formal rate-limiter. RD-F-142 n/a Storage-layout collision risk across upgrades Primary AMM contracts are immutable (V2 Factory, V3 Factory, Routers, MasterChef V1). No upgrade path = no storage collision risk. N/A by construction for the primary TVL-bearing contracts. RD-F-143 gray Reinitializable implementation (no _disableInitializers) Primary TVL-bearing contracts are immutable non-proxy deploys — reinitializer risk N/A for these. Infinity/V4 uses singleton-vault architecture. Whether any Infinity peripheral contracts use proxy pattern without _disableInitializers() cannot be confirmed without direct source inspection. Gray due to assessment gap on Infinity peripheral surface.
RD-F-138 green Hot-patch deploys without timelock (last 30 days) No evidence of hot-patch deploys outside timelock in last 30 days. Primary contracts are immutable. No recent Timelock transactions (last April 2021). No anomalous deploy events identified in profile.
RD-F-140 green Fix-merged-but-not-deployed gap No known fix-merged-but-not-deployed gap identified. BCE/USDT pool exploit (March 2025) was a third-party token mechanism issue, not a PancakeSwap protocol fix. Core AMM contracts are immutable — no patch gap possible.
RD-F-141 green Test-mode parameters in deploy No evidence of test-mode parameters in production. Factory V2/V3 are standard AMM implementations. No anomalous configuration parameters identified in verified contracts.
RD-F-144 green CREATE2 factory permits same-address redeploy No evidence of CREATE2 deployment pattern for primary contracts. Standard EVM deploys via deployer EOA. No deterministic redeployment risk identified.
RD-F-146 green New contract deploys in last 30 days No major new contract deploys identified in last 30 days (assessment date 2026-04-28). Infinity/V4 was deployed ~12 months ago (April 28 2025). GitHub last commit for pancake-v3-contracts was April 2024. No recent fresh attack surface expansion.
Cross-chain & bridge Green 17 12 of 12
RD-F-148 yellow Bridge validator count (M) LayerZero V1 does not have a traditional validator set. V1 uses Oracle (block-header submitter) + Relayer (tx-proof submitter). Effective validator count = 2 entities, but under default configuration both operated by LayerZero Labs. Not equivalent to independent multi-validator set. Functional 2-party system with single-entity default operator. RD-F-149 yellow Bridge validator threshold (k-of-M) LayerZero V1 requires both oracle AND relayer to agree for message execution (2-of-2 threshold). Structurally both must agree, but under default config both are LayerZero Labs. Not crypto-economically equivalent to independent k-of-M multisig. TrustedRemote adds application-layer whitelist but no additional validators. RD-F-150 yellow Bridge validator co-hosting Under default LayerZero V1 configuration, both oracle and relayer are operated by LayerZero Labs — maximum possible co-hosting (same entity). If PancakeSwap configured custom oracle/relayer, they could separate operators, but no evidence of custom config found for CakeProxyOFT. RD-F-156 yellow Bridge uses same key custody for >30% validators Under default LayerZero V1 configuration, effectively 100% of functional validators (oracle + relayer) are operated by LayerZero Labs. Factor threshold is >30% same custodian — this exceeds that threshold. However this reflects V1 architecture default, not PancakeSwap's specific config choice. No evidence of custom oracle/relayer separating operators. RD-F-151 n/a Bridge ecrecover checks result ≠ address(0) [★ N/A-pattern] LayerZero V1 does NOT use ecrecover for message verification. V1 security model is block-header oracle + tx-proof relayer. No ECDSA signature over cross-chain messages in this architecture. TrustedRemote uses keccak256 comparison for source contract verification. F151 (Wormhole-class ECDSA zero-address check) is structurally inapplicable. RD-F-154 n/a Default bytes32(0) acceptable as valid root [★ N/A-pattern] LayerZero V1 does NOT use Merkle root-based proof verification. No inbox contract accepting bytes32 roots. V1 proof model is oracle (block header hash) + relayer (Merkle branch proof of tx-in-block). Nomad bug class (acceptableRoot(bytes32) accepting 0) has no applicable analogue in V1 architecture. RD-F-155 n/a Bridge validator-set rotation recency LayerZero V1 oracle+relayer model does not have a discrete 'validator-set rotation' event log. Oracle is LayerZero TSS oracle / Chainlink (V1 integration). Relayer is LayerZero Labs default. Configuration changes possible via setConfig() but no specific rotation event confirmed. Cannot assess without complete V1 config history. RD-F-179 n/a LayerZero OFT DVN config (count, threshold, diversity) F179 applies to LayerZero V2 OFT adapters with DVN configuration. The CAKE OFT uses LayerZero V1 endpoint (0x3c2269811836af69497e5f486a85d7316753cf62 — confirmed V1 'Endpoint' contract, NOT V2 'EndpointV2'). V1 uses oracle+relayer model, not DVN model. DVN parameters (requiredDVNCount, optionalDVNThreshold) do not exist in V1. F179 is N/A-by-architecture. Equivalent V1 risk assessed under F148/F149. Note: if PancakeSwap migrates to V2, F179 becomes critical to reassess.
RD-F-147 green Protocol has bridge surface YES — CAKE token uses LayerZero OFT bridge. BSC ProxyOFT at 0xb274202daBA6AE180c665B4fbE59857b7c3a8091. OFT tokens on Ethereum (0x152649eA73beAb28c5b49B26eb48f7EAD6d4c898) and Base (0x3055913c90Fcc1A6CE9a358911721EEB942013a1). Also on Aptos (OtterSec-audited). has_bridge_surface=true per profile.
RD-F-152 green Bridge binds message to srcChainId LayerZero V1 trustedRemote pattern binds messages to srcChainId: trustedRemoteLookup[_srcChainId] maps each chain to expected remote address pair. Messages from unexpected chainId fail the keccak256 comparison and revert. srcChainId binding is enforced at application layer.
RD-F-153 green Bridge tracks nonce-consumed mapping LayerZero V1 endpoint tracks inboundNonce and outboundNonce per (chainId, remoteAddress) pair. Application-level failedMessages mapping stores failed cross-chain messages for retry/drop. No replay without nonce advancement. Rate-limiting caps provide additional protection.
RD-F-157 green Bridge TVL per validator ratio Ethereum CakeOFT circulating supply ~5.4M CAKE from Etherscan. At approximately $3/CAKE estimate, Ethereum bridge escrow ~$16M. With 2 effective V1 validators, per-validator ratio ~$8M — below $50M green threshold. Bridge TVL is primarily CAKE escrow, not PancakeSwap AMM TVL.
Threat intelligence & recon Red 67 8 of 8
RD-F-158 red Known-threat-actor cluster has touched protocol CONFIRMED RED. DPRK/Lazarus Group wallets (TraderTraitor / UNC4736) directly transacted with PancakeSwap's BSC swap contracts as part of laundering $263 million of the $1.46B Bybit theft (February-March 2025). PancakeSwap was the single largest DEX laundering venue in the Bybit incident — larger than SushiSwap ($74M), Curve ($47M), and Uniswap ($39M). Confirmed by: TRM Labs, Allium.so forensic analysis, FBI IC3 PSA (2025-02-26), US Senate Banking Committee letter (2025-12-15). The laundering mechanism was DeFi aggregator routing through PancakeSwap swap contracts. This constitutes confirmed interaction by a ≥1 confirmed-exploit-attributed cluster with protocol core contracts. Attribution confidence: HIGH (4 independent sources including US government). RD-F-161 red Protocol-impersonator domain registered (typosquat) CONFIRMED RED. Active confirmed typosquat domain: web2.pancake.run flagged by 14 security vendors as a live crypto drainer portal targeting PancakeSwap users (phishdestroy.io, retrieved 2026-04-28). Browser-extension impersonation attacks active. Historical precedent: hundreds of PancakeSwap typosquat domains registered Summer 2021. October 2025 Chinese X account compromise drove users to phishing sites impersonating PancakeSwap (fake 'Sir Pancake' token). Persistent high-value target: $1.73B TVL, large retail BSC user base, top-20 DeFi brand recognition. Threshold (registrar within 90 days + live drainer) clearly met. RD-F-159 gray Attacker wallet pre-strike probe (low-gas failing txs) No mempool probe pattern detected in public sources against PancakeSwap core contracts. Signal requires mempool listener infrastructure. P1 signal, not in v1 launch scope. No pre-strike test transactions against PancakeSwap contracts reported in public sources within recent monitoring period. RD-F-162 gray Known-exploit-template selector deployed by any address P2 signal requiring exploit-template DB and on-chain deploy scan. No public reporting of exploit-template-matching contracts deployed against PancakeSwap patterns as of 2026-04-28. The BCE/USDT March 2025 attack used malicious contracts targeting fee-on-transfer token interaction with the AMM — this is a token-level pattern, not a PancakeSwap-core exploit template. RD-F-163 gray Avg attacker reconnaissance time for peer-class protocols P2 analytical factor (class-average reconnaissance time). DEX-targeting attacks show variable reconnaissance periods. The BCE/USDT Mar 2025 attack appears to have been short-cycle (malicious contracts deployed for specific interaction, no documented extended reconnaissance). The 2021 DNS hijack was a direct attack. No USPD-style 78-day reconnaissance pattern reported for PancakeSwap core AMM in public sources. RD-F-164 gray Leaked credential on paste/sentry site P2 signal. No paste-site or Sentry credential leak linked to PancakeSwap infrastructure detected in public sources as of 2026-04-28. M-only curation required; no public evidence found. RD-F-165 gray Protocol social channel has scam-coordinator flag P2 signal. No specific Discord/Telegram channel admin flagged as scam-coordinator in available public sources. The October 2025 Chinese X account compromise was X-platform specific, not a Discord/Telegram admin flag. Active phishing ecosystem exists but does not rise to confirmed scam-coordinator flag on PancakeSwap's own channel admins.
RD-F-160 green GitHub malicious-dependency incident touching protocol deps No GitHub advisory flagging a malicious dependency in PancakeSwap's supply chain detected within trailing 90 days as of 2026-04-28. GitHub repo pancake-v3-contracts (last commit 2024-04-22, package_json_present: true). No npm supply chain incident reported for PancakeSwap dependencies in public security feeds.
Tooling / compiler / AI Green 7 5 of 5
RD-F-174 yellow Dependency tree uses EOL Solidity version V2 contracts compiled with Solidity 0.5.16 (last release 2020, EOL). V3 contracts compiled with Solidity 0.7.6 (last release 2020, EOL). Both V2 and V3 are immutable deployments — they cannot be recompiled. The EOL risk is limited to inability to patch bugs via recompilation, but contracts are also permanently frozen. Infinity uses 0.8.26 (supported). Yellow assigned because V2/V3 compiler is EOL even though immutability limits the practical risk.
RD-F-170 green Solc version used (known-bug versions flagged) Three distinct compiler generations: (a) V2: Solidity 0.5.16 — no high/critical known bugs per Etherscan solcbuginfo; ABIEncoderV2+Yul bugs present but V2 contracts do not use ABIEncoderV2; (b) V3: Solidity 0.7.6, optimizer 1M runs — no high-severity known bugs per published sources; (c) Infinity: solc 0.8.26 viaIR, Cancun EVM — the TransientStorageClearingHelperCollision bug affects 0.8.28-0.8.33 only; 0.8.26 predates that bug entirely. All versions free of applicable high/critical bugs.
RD-F-171 green Bytecode similarity to audited upstream with behavior deviation V3 is a confirmed Uniswap V3 fork with high bytecode similarity. Core tick-math and pool logic deviate only in fee-tier configuration, protocol-fee routing, and BSC-specific adjustments. No state-mutation ordering deviation detected — the fork was done post-BUSL expiry with a clean copy, not AI-generated. Audited by 4 firms covering V3-specific changes. Infinity is more divergent from V4 but independently audited by 3 firms.
RD-F-172 green Repo shows AI-tool co-authorship in critical files No AI-tool co-authorship (GitHub Copilot, ChatGPT Code Interpreter) detected in available commit metadata for pancake-smart-contracts, pancake-v3-contracts, or infinity-core repos. Data cache confirms no foundry.toml-based AI flags. PancakeSwap has not publicly disclosed AI-generated Solidity in production paths.
RD-F-173 green Team self-disclosure of AI-generated Solidity No public disclosure from PancakeSwap team (blog, tweet, docs) that AI-generated Solidity was used in security-critical paths. Security AMA with Hexens, Zellic, Cyfrin (June 2025) did not surface AI-tool disclosure. Team conducts traditional code reviews and formal security audits.
Response & disclosure hygiene Yellow 25 4 of 4
RD-F-176 red Disclosure SLA public No acknowledgment-time SLA published in PancakeSwap documentation or the Immunefi program description for PancakeSwap specifically. PancakeSwap's contact page lists only business partnership channels. No security@ email. No stated 72h or other ack commitment. Red: no SLA published.
RD-F-175 green Disclosure channel exists Immunefi bug bounty program active at immunefi.com/bug-bounty/pancakeswap/ — launched 2021-03-28, last updated 2026-03-16. Public, named disclosure channel. Prior confirmed use in 2021 Lottery bounty payout. Score: green.
RD-F-177 green Prior known-ignored disclosure No evidence of a known-ignored disclosure in PancakeSwap history. The 2021 Lottery bug was disclosed via Immunefi and patched within 24 hours — no exploitation occurred. The BCE/USDT 2025 incident was a novel attack vector not previously disclosed to the team. Score: green.
RD-F-178 green CVE/GHSA advisory issued against protocol No CVE or GHSA advisory found for PancakeSwap smart contracts. The BCE/USDT incident is a third-party token issue, not a PancakeSwap contract bug, and has no associated CVE. Score: green.
rubric_version v1.7.0 graded_at 2026-05-12 04:38:07 factors 184 protocol pancakeswap