Who may find this file useful
Users who understand signing risk, can protect recovery material and are willing to choose and document their recovery and telemetry settings.
Key-control & disclosure memorandum
The MetaMask memorandum examines how recovery choice, swap routing, telemetry and code licensing qualify a simple ‘self-custody’ description.
Editorial assessment: MetaMask is self-custodial, but self-custody does not mean a single recovery architecture, zero server interaction or universal code openness. Its documentation gives users meaningful choices while requiring careful reading of recovery and privacy settings.
Users who understand signing risk, can protect recovery material and are willing to choose and document their recovery and telemetry settings.
Broad dapp connectivity and recovery options improve reach and convenience while expanding the decisions and dependencies the user must secure.
Public evidence has been reviewed; account, funding, trading, withdrawal and support scenarios are specified as controlled routes with defined inputs and records.
Evidence-backed strengths
Material limitations
MetaMask is self-custodial; users authorise transactions, but service interactions remain part of some features.
Traditional Secret Recovery Phrase and social sign-in recovery use materially different architectures.
Dapp requests remain user-approved; security simulation and MetaMetrics can involve service-side data flows.
Extension and mobile repositories are public under restrictive current licences; published assessments are scoped.
Investigative analysis
Each conclusion is tied to the primary records immediately below it. Links open the publisher’s original material.
MetaMask describes itself as self-custodial because the user controls the credentials used to sign. That does not imply every feature runs without a service: routing, telemetry and security simulation can communicate with MetaMask or third parties. The useful question is which dependency is active for the chosen action.
Sources: MetaMask Support · MetaMask SupportTraditional onboarding centres on a Secret Recovery Phrase that can restore the wallet and must remain secret. Newer Google, Apple or Telegram paths use encrypted online shards and a different recovery chain. We do not label one route safer in the abstract; users need to understand the failure and account-compromise conditions of the route they select.
Sources: MetaMask Support · MetaMask SupportThe Swaps guide re-retrieved on 16 August 2026 states a 0.875% MetaMask fee, in addition to network and route-dependent costs. A flow may hide or sponsor a gas payment without eliminating all economic cost. Quote comparisons should therefore include service fee, price impact and execution route.
Sources: MetaMask SupportMetaMetrics can collect product events and selected events may be associated with wallet addresses or transaction hashes. Security simulation can send requests to MetaMask or Blockaid services. Users can manage settings, but ‘self-custody’ alone does not prove that no usage data leaves the device.
Sources: MetaMask Support · MetaMaskMetaMask maintains public extension and mobile repositories, which improves inspectability. Their current licensing terms are restrictive, so the accurate description is ‘source available for inspection’ rather than universally permissive open source. Published security reviews also apply only to the versions and scopes they name.
Sources: GitHub / MetaMask · GitHub / MetaMask · MetaMaskConsensys reported that a 2023 incident affected a third-party customer-support ticketing service and said the MetaMask browser extension and private keys were not compromised. The event remains relevant because support data can aid targeted fraud, but it should not be misreported as compromise of every wallet installation.
Sources: ConsensysClaim → evidence → gap
Each module keeps the decision consequence beside the evidence and the unresolved gap. Primary material was retrieved on .
MetaMask exposes a well-documented self-custody/signing surface.
Control stays with the user rather than a support desk.
Confirmed onchain actions may be irreversible.
RPC, simulation, swaps and MetaMetrics can create data/service dependencies.
Adjacent sources: GitHub / MetaMask · GitHub / MetaMask
Adjacent sources: MetaMask Support · MetaMask Support
Adjacent sources: MetaMask Support · MetaMask Support
Adjacent sources: MetaMask
Adjacent sources: MetaMask Support
Adjacent sources: MetaMask Support · MetaMask Support
Adjacent sources: MetaMask
Adjacent sources: MetaMask Support · MetaMask
Adjacent sources: MetaMask Support · MetaMask Support
Adjacent sources: GitHub / MetaMask · GitHub / MetaMask · Consensys
Accountable scenario file
Each row specifies a fixed protocol with defined inputs and record.
Fixed inputs: Disposable wallet, chosen recovery mode and clean device.
Record: Prompts, account restoration, imported-account boundary and settings.
PublishedFixed inputs: Controlled token/contract, approval and permit, hardware device if supported.
Record: Spender, amount, simulation, device display, reject and revoke.
PublishedFixed inputs: Same chain/pair/amount/block window with settings captured.
Record: Fee, gas, minimum received, route and service requests.
PublishedFixed inputs: No-value wallet and official support channel only.
Record: Anti-SRP warning, identity request, hand-offs and resolution scope.
PublishedChange timeline
Identity/support exposure remains distinct from key control.
Version provenance remains material.
Controlled signing evidence is specified as a fixed protocol output.
Alternatives
Broader multi-chain coverage matters more than EVM convention.
Key isolation and a smaller browser surface matter most.
Conclusion control
Method
Change log
Added ten authority/dependency claim files and four safe wallet protocols.
FAQ
Not through the traditional SRP route.
No; it isolates keys but cannot guarantee correct intent.
No; provider, route and gas costs can apply.
No; enabled services may process address, transaction, device or usage data.
Reviewed on by Candid Ledger Editorial Team; independent editorial review by Candid Ledger Review Team.