GHSA-36f9-7rg5-cpf8: Npm/@bsv/wallet-toolbox-mobile vulnerability
Reported by @echennells (Eric Chennells). Migrated from public issue #191 to a private advisory.
Affected: @bsv/wallet-toolbox / -client / -mobile. Verified in 2.1.21 and 2.1.21-parity-fix.2; the relevant code is the same at current HEAD
Summary: When createAction runs against a remote StorageClient, the storage server returns the outputs to build. buildSignableTransaction takes each non-change output's lockingScript from the storage response and signs it, without comparing it to the lockingScript the caller supplied in args.outputs. WalletPermissionsManager.createAction parses the built transaction and has args.outputs available, but uses the tx only for inputs and getFee(); it does not inspect tx.outputs. A storage provider that returns a different recipient script than requested will therefore have that script signed and broadcast, while the calling app and UI still show the originally requested recipient.
Relevant code: - signer/methods/buildSignableTransaction.js — non-change output lockingScript = asBsvSdkScript(out.lockingScript), sourced from the storage response; args.outputs is not consulted. - WalletPermissionsManager.js createAction — parses the built tx, derives spend from args.outputs satoshis + tx.getFee(), reads tx.inputs; does not read tx.outputs. - signer/methods/signAction.js, completeSignedTransaction.js — sign the as-built tx; no output comparison.
Context: Remote storage is a supported, default configuration (StorageClient over BRC-103 AuthFetch, default endpoint storage.babbage.systems, optional payment middleware). The mutual auth establishes the storage server's identity and protects the channel, but the contents it returns are not validated against the request. The attacker is the storage operator, or anyone who compromises it — not a passive network MITM. (StorageServer.processAction does compare the signed rawTx outputs to what storage stored, so a response-only MITM is rejected; an operator stores the substituted script from the start.)
Proof of concept: We ran a storage server that returns a substituted recipient output, and pointed a current yours-wallet build at it as its active storage provider. With the wallet otherwise unmodified, a payment requested to one address was built, signed, and broadcast paying a different address, the substitution was not surfaced anywhere in the wallet.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
npm/@bsv/wallet-toolbox-mobileto a version that resolves this vulnerability.Fixed in 2.4.0 - Upgrade
Upgrade
npm/@bsv/wallet-toolbox-clientto a version that resolves this vulnerability.Fixed in 2.4.0 - Upgrade
Upgrade
npm/@bsv/wallet-toolboxto a version that resolves this vulnerability.Fixed in 2.4.0
Event History
Frequently Asked Questions
Which deployments are exposed?
Deployments using @bsv/wallet-toolbox, @bsv/wallet-toolbox-client, or @bsv/wallet-toolbox-mobile with createAction against a remote StorageClient are exposed. The issue was verified in 2.1.21 and 2.1.21-parity-fix.2, and the relevant code was reported to be unchanged at current HEAD.
What does an attacker need to exploit this?
An attacker needs control of, or the ability to make malicious responses from, the remote storage provider used by StorageClient. The provider can return a different locking script for a non-change output than the script requested by the caller.
What is the practical impact of a malicious storage response?
The wallet signs and broadcasts the storage provider's substituted recipient script. The calling application and UI can still display the recipient originally supplied in args.outputs, creating a mismatch between what is shown and where funds are sent.
Can an application rely on WalletPermissionsManager to catch the substituted output?
No. WalletPermissionsManager has access to args.outputs but, according to the report, only uses the parsed transaction for inputs and fee calculation and does not inspect transaction outputs.