@whalent/agent@0.3.482
Stable wrapper for Whalent Agent core runtime
AI Security Review
scanned 22d ago · by lpm-firewall-aiSource inspection supports a warning for agent configuration mutation through runtime tools, without concrete malicious intent or an automatic install-time attack. The core handler and dispatch were inspected, although its size exceeded the citation tool's verification limit.
Static reason
One or more suspicious static signals were detected.
Trigger
A runtime tool request to install a skill or register an MCP server.
Impact
Invoked tools can persistently change agent instructions and available server capabilities. No unconsented npm install-time mutation was identified.
Mechanism
The handler downloads selected skill files and writes them into the selected scope, or inserts supplied server configuration into agent configuration files.
AI rationale
Source inspection supports a warning for agent configuration mutation through runtime tools, without concrete malicious intent or an automatic install-time attack. The core handler and dispatch were inspected, although its size exceeded the citation tool's verification limit.
Evidence
dist/mcp.cjsdist/core.cjs.claude.cursorconfig.toml
Decision evidence
public snapshotAI called this Suspicious at 83.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The skill tool downloads and installs a requested skill, defaulting to machine scope.
- Machine scope resolves beneath the user's agent platform directory, including .claude and .cursor.
- Downloaded skill files are written after a path containment check.
- The MCP registration tool can modify agent configuration, including config.toml.
- The core runtime dispatches MCP tool requests to a matching skill installation handler.
Evidence against
- package.json has no preinstall, install, or postinstall hook and no runtime self-dependency.
- Skill installation requires an install action and a skill identifier; no covert installation was established.
- No concrete credential theft or malicious payload execution was established.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStringsMinified
Source & flagged code
7 flagged · loading sourcedist/mcp.cjsView file
1#!/usr/bin/env node
L2: 'use strict';var a0_0x17fa7e=a0_0x49fb;(function(_0x1e995a,_0x1cd5f3){var _0x4cea3a=a0_0x49fb,_0x388594=_0x1e995a();while(!![]){try{var _0x5a8a14=parseInt(_0x4cea3a(0xd03))/0x1+par...
L3: || (${_0x27a557} == "string" && ${_0xcee33} && ${_0xcee33} == +${_0xcee33})`)['assign'](_0x368f64,(0x0,_0x51c059['_'])`+${_0xcee33}`);return;case _0x3207bf(0x20b5):_0x5c8688[_0x320...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•stage = ast_semantic_analysis; reason = ast_path_work_budget_exceeded; limitedFiles = 1
High
Semantic Analysis Limited
A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/mcp.cjsView on unpkg4|| (${_0x27a557} === "string" && ${_0xcee33} && ${_0xcee33} == +${_0xcee33} && !(${_0xcee33} % 1))`)['assign'](_0x368f64,(0x0,_0x51c059['_'])`+${_0xcee33}`);return;case _0x3207bf(0...
L5: || ${_0x27a557} === "boolean" || ${_0xcee33} === null`)[_0x3207bf(0x22c8)](_0x368f64,(0x0,_0x51c059['_'])`[${_0xcee33}]`);}}}function _0x75e093({gen:_0x14b8c5,parentData:_0x5b970b,...
L6: missingProperty: ${_0x1fcf09},
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@0.3.482.
See all versions of @whalent/agent for other recorded verdicts.
- Review the evidence and your use of @whalent/agent@0.3.482 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.