@whalent/agent-core@0.3.484
Core runtime for Whalent Agent
AI Security Review
scanned 22d ago · by lpm-firewall-aiExplicit runtime actions can modify agent skills and MCP configuration, warranting a behavior-specific warning under the supplied policy. No block-eligible attack chain was established; the main bundle also exceeds the citation tool's verification limit.
Static reason
One or more suspicious static signals were detected.
Trigger
Explicit MCP tool install, register, uninstall, or unregister actions.
Impact
Invoked operations can change agent instructions and the commands available to an agent.
Mechanism
Tool arguments select skills or server settings, which are passed to filesystem installation and configuration writers.
AI rationale
Explicit runtime actions can modify agent skills and MCP configuration, warranting a behavior-specific warning under the supplied policy. No block-eligible attack chain was established; the main bundle also exceeds the citation tool's verification limit.
Evidence
dist/mcp.cjs.claude.json.claude.agentsconfig.toml
Decision evidence
public snapshotAI called this Suspicious at 88.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- package.json exposes a CLI and has no npm install hooks or runtime self-dependency.
- The MCP tool handler accepts registration requests and passes caller-selected commands, arguments, and environment settings to agent configuration writers.
- The Claude configuration writer persists changes to .claude.json.
- An explicit install action downloads skill files and installs them for the selected provider and scope.
- Skill destinations include .claude and .agents directories.
Evidence against
- Configuration changes are activated through runtime tool actions; inspected source does not establish unconsented npm installation mutation.
- No concrete malicious exfiltration or hidden payload execution was established.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsMinifiedObfuscated
Source & flagged code
7 flagged · loading sourcedist/mcp.cjsView file
1#!/usr/bin/env node
L2: 'use strict';var a0_0x8f5338=a0_0x35d1;(function(_0x284375,_0x1789a7){var _0x5ab2c6=a0_0x35d1,_0x12d561=_0x284375();while(!![]){try{var _0x3711f8=-parseInt(_0x5ab2c6(0xe87))/0x1*(p...
L3: || (${_0x26c788} == "string" && ${_0x3129b1} && ${_0x3129b1} == +${_0x3129b1})`)['assign'](_0x1f5f88,(0x0,_0x54513b['_'])`+${_0x3129b1}`);return;case'integer':_0x563dac['elseIf']((...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•stage = ast_semantic_analysis; reason = ast_path_work_budget_exceeded; limitedFiles = 1
High
Semantic Analysis Limited
A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/mcp.cjsView on unpkg4|| (${_0x26c788} === "string" && ${_0x3129b1} && ${_0x3129b1} == +${_0x3129b1} && !(${_0x3129b1} % 1))`)[_0x1c15ff(0x1118)](_0x1f5f88,(0x0,_0x54513b['_'])`+${_0x3129b1}`);return;ca...
L5: || ${_0x26c788} === "boolean" || ${_0x3129b1} === null`)['assign'](_0x1f5f88,(0x0,_0x54513b['_'])`[${_0x3129b1}]`);}}}function _0x4ac92c({gen:_0x275df3,parentData:_0x15b2fd,parentD...
L6: missingProperty: ${_0x30e781},
Medium
Findings
3 High7 Medium5 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/index.cjs
MediumOversized Cli Entrypointdist/index.cjs
MediumStructural Risk Force Deep Review
LowEvaldist/mcp.cjs
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @whalent/agent-core@0.3.484.
See all versions of @whalent/agent-core for other recorded verdicts.
- Review the evidence and your use of @whalent/agent-core@0.3.484 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.