@whalent/agent@0.3.475
Stable wrapper for Whalent Agent core runtime
AI Security Review
scanned 24d ago · by lpm-firewall-aiNo concrete malicious execution or install-time attack was proven. Obfuscation combined with AI-agent configuration mutation is a dangerous capability that cannot be fully assessed from the shipped bundle.
Static reason
One or more suspicious static signals were detected.
Trigger
Invocation of the MCP or related agent configuration functionality.
Impact
This can alter an AI-agent configuration when the relevant functionality is invoked; no unconsented installation-time mutation was found.
Mechanism
A configuration helper ensures a directory exists, reads config.toml, and updates an MCP section.
AI rationale
No concrete malicious execution or install-time attack was proven. Obfuscation combined with AI-agent configuration mutation is a dangerous capability that cannot be fully assessed from the shipped bundle.
Evidence
dist/mcp.cjs
Decision evidence
public snapshotAI called this Suspicious at 82.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The package metadata has build and publish scripts but no install lifecycle hook.
- The MCP bundle is deliberately obfuscated, limiting static accountability.
- The MCP bundle contains a helper that creates and rewrites a Codex configuration file.
Evidence against
- No install-time execution was declared.
- No confirmed credential export, destructive action, or external attack endpoint was identified in the inspected code.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsMinifiedObfuscatedUrlStrings
NoLicense
Oversized source lightweight scan
Source & flagged code
8 flagged · loading sourcedist/mcp.cjsView file
4|| (${_0x729cdf} === "string" && ${_0x332576} && ${_0x332576} == +${_0x332576} && !(${_0x332576} % 1))`)['assign'](_0x133ee2,(0x0,_0x148d3b['_'])`+${_0x332576}`);return;case'boolea...
L5: || ${_0x729cdf} === "boolean" || ${_0x332576} === null`)[_0x5e3e77(0xb43)](_0x133ee2,(0x0,_0x148d3b['_'])`[${_0x332576}]`);}}}function _0x363c62({gen:_0x343c7d,parentData:_0x3f3655...
L6: missingProperty: ${_0x273bc5},
L7: depsCount: ${_0x5aa987},
L8: deps: ${_0xbb8bb4}}`};var _0x5b216c={'keyword':'dependencies','type':_0x713077(0xcfc),'schemaType':'object','error':_0x485c15['error'],'code'(_0x41ad82){const [_0x41fe79,_0x338092]...
High
Same File Env Network Execution
A single source file combines environment access, network access, and code or shell execution; review context before blocking.
dist/mcp.cjsView on unpkg · L41#!/usr/bin/env node
L2: 'use strict';var a0_0x4588e4=a0_0x2ddf;(function(_0x56a87d,_0x40dca7){var _0x8e1155=a0_0x2ddf,_0x14bf76=_0x56a87d();while(!![]){try{var _0x3c874c=-parseInt(_0x8e1155(0x1206))/0x1+p...
L3: || (${_0x729cdf} == "string" && ${_0x332576} && ${_0x332576} == +${_0x332576})`)['assign'](_0x133ee2,(0x0,_0x148d3b['_'])`+${_0x332576}`);return;case _0x5e3e77(0x259):_0x353aaf[_0x...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•
Findings
4 High7 Medium7 Low
HighSame File Env Network Executiondist/mcp.cjs
HighObfuscated Payload Loaderdist/mcp.cjs
HighObfuscated
HighSemantic Analysis Limiteddist/mcp.cjs
MediumDynamic Requiredist/mcp.cjs
MediumNetwork
MediumEnvironment Vars
MediumShips Build Helperdist/jupyter_sidecar.py
MediumOversized Source Filedist/core.cjs
MediumOversized Cli Entrypointdist/core.cjs
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowEvaldist/mcp.cjs
LowFilesystem
Affected versions and remediation
This report applies to @whalent/agent@0.3.475.
See all versions of @whalent/agent for other recorded verdicts.
- Review the evidence and your use of @whalent/agent@0.3.475 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: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.