@whalent/agent@0.3.481
Stable wrapper for Whalent Agent core runtime
AI Security Review
scanned 22d ago · by lpm-firewall-aiThe inspected source supports a warning for explicit runtime agent configuration mutation. It does not support blocking the package as malicious.
Static reason
One or more suspicious static signals were detected.
Trigger
Invoking a skill installation or MCP registration action through the running MCP server.
Impact
Changes can influence later agent behavior and configure commands that the target agent may execute.
Mechanism
Tool arguments select skills or command definitions, which are written into provider-specific skill directories or MCP configuration.
AI rationale
The inspected source supports a warning for explicit runtime agent configuration mutation. It does not support blocking the package as malicious.
Evidence
dist/mcp.cjs.claude.jsonconfig.tomlopencode.json
Decision evidence
public snapshotAI called this Suspicious at 88.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- package.json has no preinstall, install, or postinstall hook and no runtime self-dependency.
- dist/mcp.cjs resolves skill destinations in workspace or machine-level agent directories and writes downloaded files.
- The MCP tool handler registers caller-supplied commands, arguments, and environment settings in the selected provider configuration.
- Claude registration persists MCP server entries through writeClaudeConfig.
- Agent configuration mutation is exposed through the whalent_manage_mcp tool with explicit action parameters.
Evidence against
- No concrete covert exfiltration or malicious install-time mutation was established.
- Configuration changes require runtime tool invocation; the inspected code does not establish an automatic npm installation trigger.
Behavioral surface
CryptoDynamicRequireEnvironmentVarsEvalFilesystemNetwork
HighEntropyStrings
Source & flagged code
7 flagged · loading sourcedist/mcp.cjsView file
1#!/usr/bin/env node
L2: 'use strict';var a0_0x2b9281=a0_0x594f;(function(_0x4a2bf9,_0x58085e){var _0x4452cd=a0_0x594f,_0xe276a1=_0x4a2bf9();while(!![]){try{var _0x40490c=-parseInt(_0x4452cd(0x971))/0x1+pa...
L3: || (${_0x3e4101} == "string" && ${_0xf9794d} && ${_0xf9794d} == +${_0xf9794d})`)['assign'](_0x4789b2,(0x0,_0x17d467['_'])`+${_0xf9794d}`);return;case'integer':_0x54cc2d[_0x329b96(0...
High
Obfuscated Payload Loader
Source contains an obfuscated payload loader that reconstructs and executes hidden code.
dist/mcp.cjsView on unpkg · L1•stage = ast_semantic_analysis; reason = ast_path_work_budget_exceeded; limitedFiles = 1
High
Semantic Analysis Limited
A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.
dist/mcp.cjsView on unpkg4|| (${_0x3e4101} === "string" && ${_0xf9794d} && ${_0xf9794d} == +${_0xf9794d} && !(${_0xf9794d} % 1))`)['assign'](_0x4789b2,(0x0,_0x17d467['_'])`+${_0xf9794d}`);return;case'boolea...
L5: || ${_0x3e4101} === "boolean" || ${_0xf9794d} === null`)[_0x329b96(0x1cb7)](_0x4789b2,(0x0,_0x17d467['_'])`[${_0xf9794d}]`);}}}function _0x24e0cc({gen:_0x435de9,parentData:_0x85a5f...
L6: missingProperty: ${_0x525ee9},
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@0.3.481.
See all versions of @whalent/agent for other recorded verdicts.
- Review the evidence and your use of @whalent/agent@0.3.481 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.