@whalent/agent@0.3.511
Stable wrapper for Whalent Agent core runtime
AI Security Review
scanned 7d ago · by lpm-firewall-aiStatic inspection found no npm install hook, but the runtime offers a reachable tool that can persist an arbitrary MCP server command in agent configuration without a separate confirmation guard in the handler. This is a concrete dual-use agent capability, so warn rather than block.
Static reason
One or more suspicious static signals were detected.
Trigger
An MCP client invokes the server-registration action.
Impact
The configured agent may load and run the registered MCP server on a later start, extending its local capabilities.
Mechanism
The handler passes the supplied command, arguments, and environment to the selected agent configuration writer.
AI rationale
Static inspection found no npm install hook, but the runtime offers a reachable tool that can persist an arbitrary MCP server command in agent configuration without a separate confirmation guard in the handler. This is a concrete dual-use agent capability, so warn rather than block.
Evidence
dist/mcp.cjs.claude.json
Decision evidence
public snapshotAI called this Suspicious at 83.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- dist/mcp.cjs contains platform-specific skill directory resolution, consistent with an explicit skill-management feature.
- dist/mcp.cjs exposes an MCP server registration action that accepts a server name, command, arguments, and environment values.
- When registration is requested, dist/mcp.cjs writes the supplied server configuration into the selected agent configuration; for Claude Code, the file is .claude.json.
- dist/mcp.cjs routes incoming MCP tool requests to the tool handler, making the configuration mutation reachable through a tool call.
Evidence against
- package.json shows a prepublishOnly build script but no preinstall, install, or postinstall hook; the configuration mutation is tied to an explicit registration action.
- The indicated skill path resolves to platform-specific skill directories, and skill installation is handled as an explicit management action.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystem
Source & flagged code
7 flagged · loading sourcedist/mcp.cjsView file
1#!/usr/bin/env node
L2: 'use strict';var a0_0x36eba3=a0_0x59bf;(function(_0x2feb99,_0x320216){var _0x416913=a0_0x59bf,_0x4936d3=_0x2feb99();while(!![]){try{var _0x4270b2=-parseInt(_0x416913(0x63d))/0x1*(-...
L3: || (${_0x4fe319} == "string" && ${_0x75b41b} && ${_0x75b41b} == +${_0x75b41b})`)[_0x3ef7ce(0x2143)](_0x3dd5ec,(0x0,_0xa51169['_'])`+${_0x75b41b}`);return;case _0x3ef7ce(0xeb4):_0x4...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•stage = ast_semantic_analysis; reason = ast_path_work_budget_exceeded; limitedFiles = 1
High
Semantic Analysis Limited
A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/mcp.cjsView on unpkg4|| (${_0x4fe319} === "string" && ${_0x75b41b} && ${_0x75b41b} == +${_0x75b41b} && !(${_0x75b41b} % 1))`)[_0x3ef7ce(0x2143)](_0x3dd5ec,(0x0,_0xa51169['_'])`+${_0x75b41b}`);return;ca...
L5: || ${_0x4fe319} === "boolean" || ${_0x75b41b} === null`)[_0x3ef7ce(0x2143)](_0x3dd5ec,(0x0,_0xa51169['_'])`[${_0x75b41b}]`);}}}function _0x39d8ff({gen:_0x91acae,parentData:_0x537cc...
L6: missingProperty: ${_0x3dde34},
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@0.3.511.
See all versions of @whalent/agent for other recorded verdicts.
- Review the evidence and your use of @whalent/agent@0.3.511 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.