registry / twzrd-x402-gate / 0.11.2

twzrd-x402-gate@0.11.2

Security fixes in 0.11.4: the policy and paying fetch wrappers, Base Worker and strict flags could sign a payment the gate should refuse. Upgrade to 0.11.4.

Released identity 0.11.2. Buyer-side x402 trust gate: free preflight before signing USDC to a merchant. On Solana mainnet and Base mainnet (eip155:8453), a missing payTo is not approved, a buyer preflight outage does not sign unless TWZRD_FAIL_OPEN=true,

AI Security Review

scanned 2d ago · by lpm-firewall-ai

The source shows no install-time or import-time attack behavior. The flagged subprocess and network activity support the package’s explicitly invoked payment and delivery-observation features, whose limitations and data flow are described in the source.

Static reason

One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source

Trigger

A user imports the library or runs a package CLI; payment occurs only when the payment command is invoked and the precheck approves.

Impact

No malicious behavior identified. The adapter documents that its precheck does not bind the challenge ultimately signed, so it is not a challenge-bound payment firewall.

Mechanism

The safe-fetch adapter probes a payment challenge, checks it with TWZRD, and may invoke AgentCash for payment. It can then send delivery metadata to the TWZRD intelligence service.

AI rationale

Network endpoints1

intel.twzrd.xyz

Decision evidence

public snapshot

AI called this Clean at 93.0% confidence as Benign with low false-positive risk.

Evidence for AI clean decision

The only lifecycle script is prepublishOnly, which runs release verification; no install hook is declared.
The safe-fetch command identifies itself as an advisory precheck and warns that AgentCash may sign a different challenge without rechecking.
Delivery observations are derived from a completed payment; the source says response bodies are not sent.
The package exports safeFetch as a library API, while its payment flow runs only when called.

Evidence against

When invoked to pay, safe-fetch launches AgentCash through npx and passes it the current environment.
The precheck has a documented time-of-check/time-of-use gap: the later payment request may receive a different challenge.
After payment, the default delivery capture posts merchant and settlement observations to the configured TWZRD intelligence service.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetwork

Source & flagged code

5 flagged

bin/twzrd-safe-fetch.jsView file

5*/ L6: import { spawnSync } from "node:child_process"; L7: import { existsSync } from "node:fs";

High

Child Process

Package source references child process execution.

bin/twzrd-safe-fetch.jsView on unpkg · L5

bin/twzrd-gate-eval-refuse.jsView file

28const pkgRoot = join(__dirname, ".."); L29: const require = createRequire(import.meta.url); L30:

Medium

Dynamic Require

Package source references dynamic require/import behavior.

bin/twzrd-gate-eval-refuse.jsView on unpkg · L28

bin/twzrd-gate-doctor.jsView file

33// tsx resolves `src` by absolute path, so the user's cwd is safe to keep. L34: const r = spawnSync("npx", ["--yes", "tsx", src, ...process.argv.slice(2)], { L35: stdio: "inherit",

Findings

1 Critical3 High4 Medium5 Low

CriticalPrevious Version Dangerous Deltadist/safe-fetch.js

HighChild Processbin/twzrd-safe-fetch.js

HighRuntime Package Installbin/twzrd-gate-doctor.js

HighSemantic Analysis Limiteddist/evidence-verify.js#virtual:normalized:round1

MediumDynamic Requirebin/twzrd-gate-eval-refuse.js

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to twzrd-x402-gate@0.11.2.

See all versions of twzrd-x402-gate for other recorded verdicts.

Review the evidence and your use of twzrd-x402-gate@0.11.2 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-11-02 15:47:16.176Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

registry / twzrd-x402-gate / 0.11.2

twzrd-x402-gate@0.11.2

Security fixes in 0.11.4: the policy and paying fetch wrappers, Base Worker and strict flags could sign a payment the gate should refuse. Upgrade to 0.11.4.

AI Security Review

scanned 2d ago · by lpm-firewall-ai

Static reason

One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source

Trigger

A user imports the library or runs a package CLI; payment occurs only when the payment command is invoked and the precheck approves.

Impact

No malicious behavior identified. The adapter documents that its precheck does not bind the challenge ultimately signed, so it is not a challenge-bound payment firewall.

Mechanism

The safe-fetch adapter probes a payment challenge, checks it with TWZRD, and may invoke AgentCash for payment. It can then send delivery metadata to the TWZRD intelligence service.

AI rationale

Network endpoints1

intel.twzrd.xyz

Decision evidence

public snapshot

AI called this Clean at 93.0% confidence as Benign with low false-positive risk.

Evidence for AI clean decision

The only lifecycle script is prepublishOnly, which runs release verification; no install hook is declared.
The safe-fetch command identifies itself as an advisory precheck and warns that AgentCash may sign a different challenge without rechecking.
Delivery observations are derived from a completed payment; the source says response bodies are not sent.
The package exports safeFetch as a library API, while its payment flow runs only when called.

Evidence against

When invoked to pay, safe-fetch launches AgentCash through npx and passes it the current environment.
The precheck has a documented time-of-check/time-of-use gap: the later payment request may receive a different challenge.
After payment, the default delivery capture posts merchant and settlement observations to the configured TWZRD intelligence service.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetwork

Source & flagged code

5 flagged

bin/twzrd-safe-fetch.jsView file

5*/ L6: import { spawnSync } from "node:child_process"; L7: import { existsSync } from "node:fs";

High

Child Process

Package source references child process execution.

bin/twzrd-safe-fetch.jsView on unpkg · L5

bin/twzrd-gate-eval-refuse.jsView file

28const pkgRoot = join(__dirname, ".."); L29: const require = createRequire(import.meta.url); L30:

Medium

Dynamic Require

Package source references dynamic require/import behavior.

bin/twzrd-gate-eval-refuse.jsView on unpkg · L28

bin/twzrd-gate-doctor.jsView file

33// tsx resolves `src` by absolute path, so the user's cwd is safe to keep. L34: const r = spawnSync("npx", ["--yes", "tsx", src, ...process.argv.slice(2)], { L35: stdio: "inherit",

Findings

1 Critical3 High4 Medium5 Low

CriticalPrevious Version Dangerous Deltadist/safe-fetch.js

HighChild Processbin/twzrd-safe-fetch.js

HighRuntime Package Installbin/twzrd-gate-doctor.js

HighSemantic Analysis Limiteddist/evidence-verify.js#virtual:normalized:round1

MediumDynamic Requirebin/twzrd-gate-eval-refuse.js

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to twzrd-x402-gate@0.11.2.

See all versions of twzrd-x402-gate for other recorded verdicts.

Review the evidence and your use of twzrd-x402-gate@0.11.2 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-11-02 15:47:16.176Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.