@whalent/agent@0.3.477
Stable wrapper for Whalent Agent core runtime
AI Security Review
scanned 23d ago · by lpm-firewall-aiThis is a real AI-agent configuration capability whose exact caller and consent flow are obscured by the bundled code. There is no inspected evidence of an install hook, credential theft, remote payload download, or automatic configuration mutation.
Static reason
One or more suspicious static signals were detected.
Trigger
A user-invoked MCP or agent configuration operation using the package command options.
Impact
A user who applies an untrusted configuration could grant that MCP server agent-level capabilities; the inspected manifest does not show this occurring automatically during installation.
Mechanism
The bundle edits an MCP server configuration entry and persists it, allowing a configured command, arguments, environment, or URL to run through the agent platform.
AI rationale
This is a real AI-agent configuration capability whose exact caller and consent flow are obscured by the bundled code. There is no inspected evidence of an install hook, credential theft, remote payload download, or automatic configuration mutation.
Evidence
dist/mcp.cjs.claude.json
Decision evidence
public snapshotAI called this Suspicious at 84.0% confidence as Dangerous Capability with low false-positive risk.
Evidence for warning
- The manifest scripts shown are build commands, with no install lifecycle hook in the inspected section.
- The MCP bundle accepts apply and approval-related command-line options.
- The MCP bundle reads an MCP configuration and adds a server entry containing a command, arguments, optional environment, and optional URL.
- The MCP bundle writes the resulting configuration to disk.
Evidence against
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsMinifiedObfuscatedUrlStrings
NoLicense
Oversized source lightweight scan
dist/core.cjs
Source & flagged code
7 flagged · loading sourcedist/mcp.cjsView file
1#!/usr/bin/env node
L2: 'use strict';var a0_0x3b2a83=a0_0x540a;(function(_0x4a139b,_0x406cae){var _0x5d269f=a0_0x540a,_0x38a6e3=_0x4a139b();while(!![]){try{var _0x379017=parseInt(_0x5d269f(0x13cc))/0x1+-p...
L3: || (${_0x158f99} == "string" && ${_0x4ad5c4} && ${_0x4ad5c4} == +${_0x4ad5c4})`)['assign'](_0x1be106,(0x0,_0x5b47b3['_'])`+${_0x4ad5c4}`);return;case'integer':_0x1f7c3f['elseIf']((...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•stage = ast_semantic_analysis; reason = ast_path_work_budget_exceeded; limitedFiles = 1
High
Semantic Analysis Limited
A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/mcp.cjsView on unpkg4|| (${_0x158f99} === "string" && ${_0x4ad5c4} && ${_0x4ad5c4} == +${_0x4ad5c4} && !(${_0x4ad5c4} % 1))`)[_0x5a630b(0x2db)](_0x1be106,(0x0,_0x5b47b3['_'])`+${_0x4ad5c4}`);return;cas...
L5: || ${_0x158f99} === "boolean" || ${_0x4ad5c4} === null`)[_0x5a630b(0x2db)](_0x1be106,(0x0,_0x5b47b3['_'])`[${_0x4ad5c4}]`);}}}function _0x2feefe({gen:_0x1dc7e3,parentData:_0x27ebe6...
L6: missingProperty: ${_0x120c9a},
Medium
Findings
3 High7 Medium7 Low
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
LowHigh Entropy Strings
Low
Affected versions and remediation
This report applies to @whalent/agent@0.3.477.
See all versions of @whalent/agent for other recorded verdicts.
- Review the evidence and your use of @whalent/agent@0.3.477 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.