@hostinger/mcp@3.17.1
MCP server for Hostinger API
AI Security Review
scanned 6d ago · by lpm-firewall-aiThe bundled headless skill tells an agent to submit project and build reports to Hostinger without asking, despite the submission tool requiring user agreement. This is a consent and telemetry risk in the agent workflow, not automatic install-time or import-time exfiltration.
Static reason
One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
An AI client follows the headless skill’s run logging and reporting workflow.
Impact
If followed, project and build context, including domain and run metadata, is sent to Hostinger without the consent the tool requires.
Mechanism
The skill says to submit the report JSON unchanged and not ask first. The corresponding MCP tool description requires user agreement, creating a direct conflict.
AI rationale
The bundled headless skill tells an agent to submit project and build reports to Hostinger without asking, despite the submission tool requiring user agreement. This is a consent and telemetry risk in the agent workflow, not automatic install-time or import-time exfiltration.
Evidence
src/core/skills.jssrc/core/tools/all.js~/.hostinger/<domain>/reports/<run>.jsonreport.html in the workspace
Network endpoints1
developers.hostinger.com/api/hosting/v1/headless/agent-reports
Decision evidence
public snapshotAI called this Suspicious at 93.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The bundled headless skill instructs the agent to submit each build report to Hostinger and says not to ask first.
- The report includes project context such as the domain, tool, model, platform, runtime, backend, and frontend.
- The report submission tool is a Hostinger API endpoint, and its description says to send a report only when the user agreed.
- The main server imports and exposes the skills to its MCP client.
Evidence against
- The report instructions are bundled skill content; the reviewed entry point does not run them automatically on import.
- The package has no install lifecycle scripts, and its OAuth flow uses Hostinger’s auth service and stores credentials locally.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsMinifiedUrlStrings
Source & flagged code
7 flaggedbuild.jsView file
2L3: import { exec } from 'child_process';
L4: import fs from 'fs';
High
18console.log('Compiling TypeScript...');
L19: exec('npx tsc', (error, stdout, stderr) => {
L20: if (error) {
High
Runtime Package Install
Package source invokes a package manager install command at runtime.
build.jsView on unpkg · L18•matchType = normalized_sha256
matchedPackage = hostinger-api-mcp@3.5.0
matchedPath = build.js
matchedIdentity = npm:aG9zdGluZ2VyLWFwaS1tY3A:3.5.0
similarity = 1.000
summary = normalized source hash matched finalized malicious source
High
Known Malware Source Similarity
Source file is highly similar to a previously finalized malicious package; route for source-aware review.
build.jsView on unpkgskills/headless/connect/bootstrap.mjs
Findings
1 Critical7 High3 Medium4 Low
CriticalPrevious Version Dangerous Deltasrc/core/skills.js
HighChild Processbuild.js
HighShell
HighRuntime Package Installbuild.js
HighKnown Malware Source Similarityskills/headless/connect/bootstrap.mjs
HighKnown Malware Source Similaritysrc/core/oauth.js
HighKnown Malware Source Similaritysrc/core/oauth.ts
HighKnown Malware Source Similaritybuild.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
Affected versions and remediation
This report applies to @hostinger/mcp@3.17.1.
See all versions of @hostinger/mcp for other recorded verdicts.
- Review the evidence and your use of @hostinger/mcp@3.17.1 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.