registry / @blockrun/llm / 3.14.2

@blockrun/llm@3.14.2

TypeScript SDK for BlockRun - every frontier model behind one OpenAI-compatible client. Pay with a BlockRun API key or per-call in USDC. Smart routing picks the cheapest capable model. No rate limits.

AI Security Review

scanned 1mo ago · by lpm-firewall-ai

The package is not proven malware because there is no lifecycle execution or observed exfiltration. Its broad wallet-key discovery and payment capabilities are nevertheless a real dangerous runtime surface.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

Calling wallet discovery or import helpers, constructing the Anthropic-compatible client without credentials, or making a wallet-billed API request.

Impact

A consumer or agent that invokes these APIs can expose or reuse local wallet credentials and authorize USDC payments.

Mechanism

Cross-application wallet-key discovery, local key copying, and payment signing.

AI rationale

Evidence

dist/index.cjs<home>/.*/wallet.json<home>/.blockrun/.session<home>/.blockrun/.session.backup-<timestamp>

Network endpoints1

blockrun.ai/api

Decision evidence

public snapshot

AI called this Suspicious at 87.0% confidence as Dangerous Capability with low false-positive risk.

Evidence for warning

The exported wallet scanner enumerates hidden home-directory folders and reads wallet.json private keys.
An exported import function copies a discovered private key into the package wallet session file.
Constructing the Anthropic-compatible client without credentials automatically creates or loads a wallet.
Wallet payment signing is reachable through user-invoked API requests.

Evidence against

package.json has no preinstall, install, or postinstall hook.
The inspected entrypoints do not execute scanning, network requests, or shell commands merely on import.
The code does not send discovered private keys over the network; importing a discovered wallet requires an explicit function call.

Behavioral surface

Source

CryptoEnvironmentVarsFilesystemNetwork

Supply chain

HighEntropyStringsUrlStrings

ManifestNo manifest risk signals triggered.

scanned 2 file(s), 731 KB of source, external domains: api.blockrun.ai, base-rpc.publicnode.com, base.llamarpc.com, basescan.org, blockrun.ai, bridge.base.org, mainnet.base.org, pay.coinbase.com, sol.blockrun.ai, user.blockrun.ai, www.coinbase.com

Source & flagged code

5 flagged

dist/index.jsView file

Lines 66-106javascript

66 if (temperature === void 0 || temperature === null) {

67 return;

68 }

69 if (typeof temperature !== "number") {

70 throw new Error("temperature must be a number");

71 }

72 if (temperature < 0 || temperature > 2) {

73 throw new Error("temperature must be between 0 and 2");

74 }

75}

76function validateTopP(topP) {

77 if (topP === void 0 || topP === null) {

78 return;

79 }

80 if (typeof topP !== "number") {

81 throw new Error("topP must be a number");

82 }

83 if (topP < 0 || topP > 1) {

84 throw new Error("topP must be between 0 and 1");

85 }

86}

87function validatePrivateKey(key) {

88 if (typeof key !== "string") {

Critical

Wallet Drain

Source uses private key material to transfer cryptocurrency funds.

dist/index.jsView on unpkg · L86

89 throw new Error("Private key must be a string");

90 }

91 if (!key.startsWith("0x")) {

92 throw new Error("Private key must start with 0x");

93 }

94 if (key.length !== 66) {

95 throw new Error(

96 "Private key must be 66 characters (0x + 64 hexadecimal characters)"

97 );

98 }

99 if (!/^0x[0-9a-fA-F]{64}$/.test(key)) {

100 throw new Error(

101 "Private key must contain only hexadecimal characters (0-9, a-f, A-F)"

102 );

103 }

104}

105function validateApiUrl(url) {

106 let parsed;

dist/index.cjsView file

Lines 183-223javascript

183 if (temperature === void 0 || temperature === null) {

184 return;

185 }

186 if (typeof temperature !==

Source & flagged code

5 flagged · loading source

dist/index.jsView file

86} L87: function validatePrivateKey(key) { L88: if (typeof key !== "string") { ... L113: throw new Error( L114: `Invalid protocol: ${parsed.protocol}. Use http:// or https://` L115: ); ... L166: } L167: const key = options.apiKey ?? (options.privateKey === void 0 && typeof process !== "undefined" ? process.env?.BLOCKRUN_API_KEY : void 0); L168: if (key === void 0) return void 0; ... L222: try { L223: body = await response.json(); L224: } catch {

Critical

Wallet Drain

Source uses private key material to transfer cryptocurrency funds.

dist/index.jsView on unpkg · L86

dist/index.cjsView file

203Trigger-reachable chain: manifest.main -> dist/index.cjs L203: } L204: function validatePrivateKey(key) { L205: if (typeof key !== "string") { ... L230: throw new Error( L231: `Invalid protocol: ${parsed.protocol}. Use http:// or https://` L232: ); ... L283: } L284: const key = options.apiKey ?? (options.privateKey === void 0 && typeof process !== "undefined" ? process.env?.BLOCKRUN_API_KEY : void 0); L285: if (key === void 0) return void 0; ... L339: try { L340: body = await response.json(); L341: } catch {

Critical

Trigger Reachable Dangerous Capability

A package entrypoint or install-time lifecycle script reaches a source file with blocking dangerous behavior.

dist/index.cjsView on unpkg · L203

Findings

2 Critical6 Medium4 Low

CriticalWallet Draindist/index.js

CriticalTrigger Reachable Dangerous Capabilitydist/index.cjs

MediumNetwork

MediumEnvironment Vars

MediumStructural Risk Force Deep Review

MediumAi Review Evidencedist/index.cjs

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to @blockrun/llm@3.14.2.

See all versions of @blockrun/llm for other recorded verdicts.

Review the evidence and your use of @blockrun/llm@3.14.2 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: 2026-09-11 18:07:30.024Z (UTC).

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

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code