Security and compliance

Security and trust

No custody, no signing on your behalf, no promises to recover lost funds. Here is what we defend, how we prove it, and what we won't do.

Where our responsibility ends

What we don't do

No asset custody

We never hold, move or custody user assets. We don't offer custodial accounts, MPC/HSM signing or programmatic signing. Your own wallet always signs and broadcasts revocations and transactions.

No private keys, ever

Our APIs, pages, logs and support staff never ask for private keys or seed phrases. We only read public onchain data and cannot sign on your behalf.

No recovery guarantees

We do not promise to recover lost assets. At-risk amounts estimate exposure, not recoverable funds. This tool helps reduce future risk rather than reverse past losses.

Security practices

How we protect your account and data

Only key hashes are stored

Plaintext keys are not stored, and upstream credentials are kept out of application data. Key disabling, limit changes and routing changes take effect globally within 30 seconds.

Least privilege and multifactor authentication

Console permissions are role-based. Administrators and owners must enable multifactor authentication; accounts without it are restricted to read-only roles.

Two-person review and auditing

Sensitive actions such as data deletion, emergency suspension and billing-period adjustments require confirmation from a second administrator. All permission and key changes are audited with the actor, time, source IP and before-and-after values.

Revocations require an onchain state check

A broadcast is not a revocation. We read state back by transaction hash and handle pending, replaced, and reorged transactions. A failed revocation is never shown as success.

Explainable risk assessments

Each risk includes a reason, evidence, source and update time. Commercial partnerships do not influence risk assessments.

Durable usage records

Every usage event is written before it hits the ledger, and duplicates are dropped automatically. Anomalies trigger alerts and caps; charges can be recomputed and reconciled.

Security control coverage

12 attack surfaces and their controls

We suspend any feature with a critical issue: wrong signatures, wrong revocations, key exposure or commercial tampering with risk assessments. We don't ship first and fix later.

IDAttack surfaceRequired controlsRelease evidence
S-01Wallet connections and domainsHTTPS, domain protection, network and chain identifiers, and clear notices before connectingDomain and certificate inventory, and connection replay records
S-02Transaction previews and revocationsIndependent decoding of calldata, onchain simulation, raw target and outcome shown before confirmationPer-transaction comparison of simulated and onchain outcomes
S-03RPC and indexing servicesCross-checks across sources, displayed block heights, latency indicators and fallback on anomaliesNode failover and inconsistent-data drills
S-04Chain reorganizations and transaction replacementsRequired confirmations, state checks, retries after reorganizations, and no false success reportsTestnet reorganization and replacement tests
S-05Contract and token metadataContract addresses take precedence, metadata sources are ranked, and proxy and upgrade status is shownMalicious metadata sample tests
S-06Web applicationInput and output encoding, session protection, permission checks, rate limits, dependency scans and security response headersDAST, dependency inventory and authorization bypass tests
S-07Operations consoleMultifactor authentication, least privilege, two-person review, release rollback and action auditingRole matrix, audit logs and rollback drills
S-08Development and dependenciesNo committed secrets or sensitive data, code review, pinned dependencies, SBOM and release approvalReview records, dependency scans and secret scans
S-09Browser extensionLeast privilege, signed releases, version rollback, permission explanations and independent reviewsStore review records and permission tests
S-10Data and privacyData minimization, tiered access, redaction, retention periods, deletion and exportData flow diagrams, access spot checks and deletion replay records
S-11Partnerships and monetizationRisk assessments isolated from commercial teams, promotion labels and partnership exclusion listsPartnership approval and content review records
S-12Availability and incident responseRate limits, caching, graceful degradation, emergency shutdown, notifications and disaster recoveryLoad tests, failover drills and post-incident reviews
Security advisories

Vulnerability incidents and risk-label changes

Each advisory includes its source, reviewer and update time.

Compliance and data processing

Data minimization
We collect only the data the service needs. Onchain data is already public, and we don't build profiles on top of it to sell.
Tiered access
Access is tiered by role, and sensitive fields are redacted. Every access lands in the audit log.
Retention
Usage details are kept for 90 days. Audit logs and bills stay for the legally required period, then are removed automatically.
Deletion and export
Export or delete your platform records in Account and Data. Public onchain data is unaffected.
Data storage region
Data is stored in the service's region by default. Enterprise customers can agree on a data residency region.
Automatic revocation and notifications
Automatic revocation, gas subsidies, address profiling, notifications, and data export each get their own permission and privacy notice. All are off by default.
Payments and tax
Refund rules are set out in the Terms of Service. Section 3 lists the operating entity, regions, payment channels, and tax treatment.

Vulnerability disclosure process

1
Submit a reportProvide reproduction steps, affected scope and evidence. Do not disclose unpatched issues publicly.
2
Acknowledgment and triageWe acknowledge receipt within 3 business days and classify the issue by severity.
3
Remediation and reviewEvery fix needs code review and a second person's sign-off. Signing or revocation logic stays frozen until verification passes.
4
Advisory and post-incident reviewAfter deploying a fix, we publish an advisory with the affected scope and timeline and complete a post-incident review.
Do not send wallet credentials

Do not send credentials that can control assets, even for reproduction. Use a test wallet address instead.

FAQ

Security and responsibility

We can read public onchain balances, approvals and transactions, just like a block explorer. We cannot move your assets: every transfer requires your wallet's signature, and signing authority always stays with you.

Automatic revocation is off by default. We enable it only when the scope is provable and failures are recoverable. The wallet and chain must also pass security checks. Rules cover only the approval types you tick, within a budget and an expiry date. Any rule can be paused at any time.

We consider allowance size, the approved contract's reputation and history, prolonged inactivity and matches to published vulnerability incidents. Each assessment retains a reason, evidence, source and update time for individual verification.

No. Findings are produced independently of the commercial team. Partnerships and promotions are clearly labeled. An exclusion list keeps us from taking deals that would sway findings.

No, and we do not promise recovery. Onchain transactions are irreversible. We help identify and revoke remaining high-risk approvals and monitor new approval changes to reduce the risk of further loss.

Found a security issue?

Report it through our security channel with reproduction steps and impact. We acknowledge every report within 3 business days.