Base Chain is a fast, low-cost, decentralized network that delivers sub-second, sub-cent global transactions. It is the foundation for a new economy where anyone, anywhere can participate.
Live
Triaged by Immunefi
Runnable PoC Required
KYC required
Documentation
All programming languages within the explicitly listed in-scope assets are covered, subject to this policy’s exclusions.
Client source baseline: base/base v1.6.0-rc.9, commit 689054ee14933cb2ee5bd54eadcd8dacbe1cc517. Its declared base/reth dependency is 2b20e42de9c42b51ea0fb56599fda3c2f22640ef. The client pin does not automatically include every dependency/component or finalize the other repositories' scope. Deployed contracts in scope, below, lists each contract's address, implementation, version and source revision.
Networks and code in scope
Everest will not be live on Sepolia during the competition; it is tested on the designated local Everest devnet built from the pinned commit (just devnet up everest). The Base-native contracts deployed on Base are in scope: AggregateVerifier, TEEVerifier, DisputeGameFactory, AnchorStateRegistry, TEEProverRegistry, ProtocolVersions, NitroValidator, CertManager and P384Verifier (github.com/base/nitro-validator at 0ea0d123). Contracts inherited from Optimism that Base does not deploy are out of scope even though they appear in base/contracts: SuperFaultDisputeGame and super-root game types, OPSuccinctFaultDisputeGame and its AccessManager, OPContractsManager. The legacy fault-proof contracts still deployed on Ethereum Sepolia (FaultDisputeGame, PermissionedDisputeGame, MIPS, PreimageOracle and their DelayedWETH proxies) are also out of scope.
Everest scope covers Base Transactions (EIP-8430, type 0x79) and gas-limit billing at the immutable base/base commit in Assets in Scope. This includes transaction encoding, authorization, replay protection, execution, settlement, native nonce/context precompiles, payer RPC when enabled, admission, block building, receipts, batch encoding/derivation and the proof-boot Everest gate. Use the Base-transactions/base-tx names at the recorded rc.9 pin. Historical EIP-8130 names do not exclude retained functionality.
Gas-limit billing is in scope for standard user transactions and Base Transactions when it is enabled in the Everest environment, including execution and proving consistency, settlement, builder reservations, and receipt and header accounting.
OP-derived contracts that Base deployed are in scope as listed in Deployed contracts in scope, below: OptimismPortal2, SystemConfig, DisputeGameFactory, L2ToL1MessagePasser, L1Block, the fee vaults, the bridge and messenger contracts and their factories, and the listed L2 predeploys.
The pinned code defines the implementation under review. This policy defines scope, intended security requirements and eligible impacts. Code takes precedence over the EIP and documentation for implementation details, but an implementation bug is not intended behavior merely because it exists in the code. A discrepancy without an in-scope impact is not eligible for a reward.
The in-scope deployment pipeline is base/contract-deployments. Deployment scripts inside base/contracts, such as those that wire MockVerifier, are test tooling and out of scope. Legacy endpoints that are no longer provisioned, such as the authenticated websocket-proxy endpoint, are out of scope.
Deployed contracts in scope
The tables below list the contracts deployed on Ethereum Sepolia (chain ID 11155111) and Base Sepolia (chain ID 84532), and the native contracts of base/base. Addresses, implementations and versions were read on 16 November 2026 at Ethereum Sepolia block 12,149,624 and Base Sepolia block 59,691,356. These deployments run earlier revisions than the pinned commit and do not include Everest. They matter only for findings that depend on deployed pre-Everest state, tested on a local fork (see PoC Requirements). Source revisions: base/contracts v8.4.0 (8c21f6d0) for the Denim, 20 October and 4 November deployments, and v8.3.0 (385f21a4) for the Cobalt deployment; base/nitro-validator 0ea0d123; deployment records in base/contract-deployments e6797dfd. Exceptions, where the deployed code is not one of these revisions: AnchorStateRegistry runs v8.0.0 (ddd5a09e) plus the patch https://github.com/immunefi-team/audit-comp-base-everest-contract-deployments/blob/e6797dfd6ac03b35d0ddd7eb369ab94bbc5fcd31/archive/legacy/sepolia/2026-04-20-activate-multiproof/patch/asr-reset-anchor-game.patch from the 20 April 2026 multiproof activation, and still reports 3.7.0; DelayedWETH (Multiproof) runs v8.0.0 (ddd5a09e) from the same deployment, without a patch; SystemConfig runs v8.3.0 (385f21a4) plus the patch https://github.com/immunefi-team/audit-comp-base-everest-contract-deployments/blob/e6797dfd6ac03b35d0ddd7eb369ab94bbc5fcd31/archive/evm/2026-09-14-cobalt-upgrade/patch/max-gas-limit.patchapplied to the listed Sepolia deployment, hence its version 3.14.0+max-gas-limit-2000M. Everest uses the native precompiles of the designated local Everest devnet, built from the competition commit. At that commit, the devnet deploys its L1 contracts from the artifacts bundled with its setup tooling, not from base/contracts v8.5.0-rc.1. Base fixed this on main in base/base#6680 (base/base#6680). To use it, apply base/base#6682 and then base/base#6680 on top of the competition commit, in that order (base/base#6680 does not apply on its own); just devnet up everest then builds the L1 contracts and L2 predeploys from base/contracts v8.5.0-rc.1 (b9b76d97). Both change only the devnet setup, not in-scope code. The Everest devnet still runs no proposer, prover or challenger; proof components follow the proof-component PoC exception.
Ethereum Sepolia (L1)
Base Sepolia (L2)
Native contracts (base/base)
| Contract | Address | Implementation | In scope |
|---|---|---|---|
| B20Factory | 0xB20f000000000000000000000000000000000000 | Native precompile in base/base at the competition commit | Yes |
| ActivationRegistry | 0x8453000000000000000000000000000000000001 | Native precompile in base/base at the competition commit | Yes |
| PolicyRegistry | 0x8453000000000000000000000000000000000002 | Native precompile in base/base at the competition commit | Yes |
| B20 tokens (asset and stablecoin variants) | Created by B20Factory | Native precompile in base/base at the competition commit | Yes |
| Nonce Manager | 0x813000000000000000000000000000000000aa01 | Native precompile in base/base at the competition commit | Yes |
| Transaction Context | 0x813000000000000000000000000000000000aa02 | Native precompile in base/base at the competition commit | Yes |
Production configuration
The values below describe the typical Base setup after Everest: the configuration Base runs today with Denim, which it expects to keep when Everest activates. PoCs use this configuration on the designated local Everest devnet (see Everest environment, below, and PoC Requirements). Lines marked “On Sepolia” describe the pre-Everest deployments and apply only to findings tested on a local fork. They are facts for this competition, even where the configuration files sit in folders that are otherwise out of scope, such as base/base/etc. Findings that only appear under a different configuration are out of scope unless the report shows an unprivileged way to reach it.
- Operator fee: operatorFeeScalar = operatorFeeConstant = 0.
- Resource metering is an optional optimization, off by default, and fails open by design. The local devnet turns it on for base-builder (--builder.enable-resource-metering in etc/docker/docker-compose.yml); remove that flag when preparing the competition test environment. This is a testing-configuration override, not a change to the audited implementation. Base made the same change on main in base/base#6682, which also applies cleanly to the competition commit.
- Bundles are not live and are not a private order-flow mechanism.
- The following experimental or testing modes are excluded from the competition’s scope:
- Shadow indexing: --enable-shadow-indexer.
- Shadow sequencing: --sequencer.shadow-blocks-per-cycle and --sequencer.shadow-funding-*.
- Consensus development RPC: --rpc.dev-enabled.
- Dangerous L1 overrides: overrides of the DA batch inbox or batcher sender.
- Deferred validity-signature mode: --validity-signature-mode.
- Experimental Reth synchronization and networking: --snap.v2 and --net-if.experimental.
- Research-only transaction limits: --pooled-tx-response-soft-limit and --pooled-tx-pack-soft-limit.
- Reth development and testing modes: --dev*, --testing.* and the testing_buildBlockV1 RPC.
- Reth debug CLI modes: --debug., including options that skip state-root or genesis validation, simulate reorgs or consensus, or skip fork-choice updates or new-payload processing. These CLI modes are distinct from debug builds and debug_ RPC methods.
- Experimental engine-tuning options: --engine.txpool-prewarming, --engine.share-sparse-trie-with-payload-builder, --engine.suppress-persistence-during-build, --engine.prewarm-reuse, --engine.state-root-task-compare-updates, --engine.disable-sparse-trie-cache-pruning and --engine.disable-bal-*.
- Payer relay on RPC nodes: --rollup.payer-enabled / --rollup.payer serve payer_getTerms and payer_sendTransaction. Off by default and not set anywhere.
These exclusions apply only to findings that require the listed unsupported modes or settings. Shared code reachable under the supported competition configuration remains in scope. Using an approved local test harness does not itself make a finding out of scope.
- Base batcher’s enabled shadow mode (BASE_BATCHER_SHADOW_ENABLED) remains in scope. It is distinct from the excluded shadow-indexer and shadow-sequencer modes
- Block production: on the designated local Everest devnet, base-builder runs as the sequencer and, after Denim, builds each block once with the native payload builder; base-client follows it and forwards transactions to it. Batching: op-batcher v1.16.5 with Base Transaction support (base/optimism op-batcher/v1.16.5-basetx.2) posts singular batches; base-batcher runs in shadow mode with its own key and a separate shadow inbox.
- Cancun and later hardforks are active; pre-Cancun code paths are unreachable.
- AggregateVerifier 0.3.0 has a slow interval of 600/30 and a fast interval of 6000/300 (20 intermediate roots each), chosen per game from its starting block. The fast interval starts at ProtocolVersions index 13 (Denim); Denim is active, so PoCs use 6000/300.
- ZK verifier: AggregateVerifier uses ZkVerifier; on Sepolia, version 0.1.0 at 0xa738E13d97A3f10b307C2CD190edd3E227A72D68.
- TEE prover fleet: 4 instances, each with 39.06 GiB (40,000 MiB) of enclave memory. This has been Base’s typical setup since Base Azul and is expected for Everest; Base may run fewer instances at low load and scales up (as high as 16) to recover after downtime.
- On Sepolia, OptimismPortal2.proofMaturityDelaySeconds = 0 and AnchorStateRegistry.disputeGameFinalityDelaySeconds = 0: a withdrawal can be finalized in the block after its game resolves. This applies only to findings tested on a local fork of Sepolia. PoCs use the Base mainnet values: both delays are 1 day, with the dispute windows in Assumptions about Base’s response (5 days with one accepted proof; 1 day after a second proof of the other type).
- Everest environment (the designated local Everest devnet, just devnet up everest: Azul, Beryl, Cobalt and Denim activate at L2 blocks 20, 21, 22 and 25, and Everest at block 100):
- Gas-limit billing is active: every non-deposit transaction is billed for its gas limit, plus the payer authentication cost for a sponsored Base Transaction.
- Keystore code is removed at the competition commit: only secp256k1 sender and payer authentication remain, and delegation (type 0x01) is the only account change, at most one per transaction. Zenith is an empty, genesis-only upgrade that gates no execution behavior, and proof boot rejects it.
- The payer RPC (payer_getTerms, payer_sendTransaction) is disabled by default.
- Mempool ordering by declared coinbase tip divided by gas_limit; per-account limits of 16 per payer (a self-paying sender counts as its own payer) and 64 per allowlisted payer. The per-sender limit and the --rollup.mempool-sender-limit and --rollup.mempool-trusted-delegation-targets flags were removed.
- Base cost of 12,000 gas; payer authentication is a fixed 5,100 gas (secp256k1 recovery 3,000 plus one cold read 2,100), billed on top of gas_limit for a sponsored transaction; MAX_CALL_PHASES_PER_TX = 1,024; MAX_ACCOUNT_CHANGES_PER_TX = 1.
- Nonce-free window of 30,000 ms at inclusion and 20,000 ms in the mempool; replay ring of 340,000 entries.
- Signature and replay domains: 0x7900 (sender), 0x7901 (payer) and 0x7902 (replay_id).
- Execution client dependency: github.com/base/reth at 2b20e42de9c42b51ea0fb56599fda3c2f22640ef, as recorded in the deployment manifest.
Build Commands, Test Commands, and How to Run Them
Detailed build commands, test commands and run instructions are in the README files and documentation of each repository:
- Offchain components: https://github.com/immunefi-team/audit-comp-base-everest-base
- Deployment and upgrade scripts: https://github.com/immunefi-team/audit-comp-base-everest-contract-deployments
- Implementation contracts: https://github.com/immunefi-team/audit-comp-base-everest-contracts
- TEE attestation contracts: https://github.com/immunefi-team/audit-comp-base-everest-nitro-validator
Asset Accuracy Assurance
Bugs found on assets incorrectly listed in-scope are valid.
Code Update Assurance
Duplicate submissions of bugs are valid until the relevant bug fix is public. Once a fix is public, the bug is out of scope and later submissions of it are invalid.
A fix is public when it is listed in the Mid-Contest Changelog on this page with its UTC timestamp. Base adds every public PR or commit that fixes a security issue to the changelog within 24 hours of it becoming public, and announces it in the competition Discord channel. Reports submitted before the changelog timestamp are eligible, including as duplicates. Reports submitted after it are invalid.
A fix or advisory published after a submission does not, by itself, establish that the report is valid or set its severity.
The project commits to keeping all information about bug findings private until each fix is disclosed through the process described in the Bug Fix Policy below. This includes bug findings found independently by the project or from concurrent private audits.
Any necessary change must be documented, timestamped and communicated by Base and Immunefi. Changes must not retroactively invalidate eligible submissions or reduce accrued eligibility; define prospective treatment and any researcher-beneficial clarification explicitly.
Code Freeze Assurance
The commits listed in Assets in Scope are the competition baseline. The codebase is not frozen: Base may apply fixes during the competition under the Bug Fix Policy in Impact Terms.
Mid-Contest Changelog
Each entry lists the UTC timestamp, the PR or commit, and the issue it fixes. No entries yet.
KYC Requirement
Researcher KYC is required for payouts and is intended to be collected and processed by Immunefi or its disclosed KYC provider, not delivered to Coinbase. The following information will be required:
- Full name
- Date of birth
- Proof of address (either a redacted bank statement with address or a recent utility bill)
- Copy of Passport or other Government issued ID
Security researchers are required to submit KYC within 7 days of KYC being requested, else their rewards may be forfeited. Immunefi may make exceptions due to extenuating circumstances.
For the full process, see KYC Process for Audit Competitions (link: https://immunefisupport.zendesk.com/hc/en-us/articles/32264466580497).
Standardized Rules
This competition follows Immunefi's standard competition rules. For the full default rules on validity, duplicates, known issues, severity, disputes, KYC, payments, and publication, read How Audit Competitions Work: Rules and Policies. Where this program page defines a specific rule, the program page prevails.
Primacy of Impact vs Primacy of Rules
Base adheres to the Primacy of Rules, meaning the whole program is run strictly under the terms and conditions stated on this page.
Insights
Insight reports are not accepted in this competition. Every report must demonstrate an in-scope impact with a runnable PoC; reports that only describe best practices, code quality, documentation differences or theoretical concerns are closed.
Duplicates and repeated reports
Base never closes a duplicate of a valid report. It marks the report as a duplicate of the earliest one, and Immunefi applies the duplicate reward rule. A later report of the same root cause by the same researcher is closed in favour of their first report. Each report must cover one root cause; triage splits reports that mix several. All reports of one root cause receive the same classification.
Dispute Resolution
Disputes may only address report validity, duplicate determinations, and severity.
If a researcher disagrees with a closure decision while the competition is live, they may request mediation. A mediation fee, set by Immunefi, is required to initiate the review.
After the evaluation phase, a dispute period opens for appealing decisions. Immunefi announces a final 48-hour deadline for the last disputes. Once the window closes, results are final.
Immunefi has final say on validity, duplicates, and severity, based on the terms of this program.
Base replies to each mediation request within 5 business days.
Responsible Publication Policy
Immunefi will publish bug reports, earnings, and a leaderboard for this Audit Competition.
Security Researchers may publish their bug reports as well, but only after Immunefi has published the valid bug reports as part of the competition results.
Eligibility Criteria Security researchers who wish to participate must adhere to the rules of engagement set forth in this program and cannot be:
- On OFACs SDN list
- Official contributor, both past or present
- Employees and/or individuals closely associated with the project
- Security auditors that directly or indirectly participated in an audit review of the code in scope (Such auditors may still participate in this program only if they receive project permission)
- From a restricted country or territory per Immunefi's Terms of Use (link: https://immunefi.com/terms-of-use/)
- Under 18 years of age. Participants under 18 cannot pass KYC and will not receive rewards.
Submission Fees
A submission fee, set and managed by Immunefi, applies to each report and is shown at submission time. If a researcher disagrees with a closure decision, initiating a mediation also requires a fee set by Immunefi.


