registry / mcp-scraper / 0.66.6
mcp-scraper@0.66.6
Upgrade to mcp-scraper 0.98.0 or newer. Older releases include obsolete local execution code; see the packaged README for hosted-client migration.
MCP server for MCP Scraper web intelligence tools
AI Security Review
scanned 2mo ago · by lpm-firewall-aiInspected source shows a commercial MCP web-intelligence client and hosted API with no registry install hooks and no unconsented foreign agent writes. High static hits are package-aligned dual-use features (hosted ChatGPT/Claude capture, vendor LLMs, ffmpeg, explicit claude mcp add).
Static reason
One or more suspicious static signals were detected.
Trigger
A consumer would have to run a bin such as mcp-scraper, mcp-scraper-cli, or paa-api after install.
Impact
No unconsented install-time mutation, secret theft, or foreign agent hijack was shown.
Mechanism
User-invoked MCP client and optional first-party agent config helper
AI rationale
Inspected source shows a commercial MCP web-intelligence client and hosted API with no registry install hooks and no unconsented foreign agent writes. High static hits are package-aligned dual-use features (hosted ChatGPT/Claude capture, vendor LLMs, ffmpeg, explicit claude mcp add).
Decision evidence
public snapshotAI called this Clean at 84.0% confidence as Benign with medium false-positive risk.
Evidence for AI clean decision
- package.json has no preinstall, install, or postinstall scripts; published entrypoints are user-invoked bins under dist.
- The stdio MCP server reads a user API key and calls the product host https://mcpscraper.dev by default.
- Claude Code MCP registration runs only when the user passes --apply on mcp-scraper-cli agent install, and it upserts the mcp-scraper server only.
- ChatGPT and Claude browser capture is an explicit query_fanout_workflow on a user-opened hosted session, not silent account takeover.
- DeepInfra and OpenRouter requests use first-party vendor API keys for package LLM work, not a hidden consumer-key siphon.
- ipify and related IP lookups run inside a harvest debug browser page for proxy geolocation.
Evidence against
- The tarball ships a large hosted API server with ffmpeg spawn, OAuth redirect allowlisting, Stripe, and browser automation.
- mcp-scraper-cli can spawn the claude CLI to change user-scope MCP config when the user asks.
- Static high labels fired on those dual-use product surfaces because the bundled server is oversized.
Behavioral surface
Source & flagged code
10 flaggeddist/server-5WPS3LZD.jsView file
9163// src/services/site-architecture-auditor/python/graph-runner.ts
L9164: import * as childProcess from "child_process";
L9165: import * as path from "path";
High
Child Process
Package source references child process execution.
dist/server-5WPS3LZD.jsView on unpkg · L9163252import {
L253: isPrivateIpAddress,
L254: loadHtmlDocument,
...
L480: ctaHeadingItalic: "real data.",
L481: ctaBody: "MCP Scraper gives your AI workflows the web intelligence they need \u2014 SERP data, People Also Ask harvests, page extraction, YouTube transcripts, and more. All via API...
L482: sections: [
...
L743: answer: `<strong>ChatGPT makes things up because its RLHF training consistently rewarded fluent, complete-sounding answers \u2014 and human raters often cannot tell in the moment w...
L744: source: "https://techcrunch.com/2026/05/05/openai-releases-gpt-5-5-instant-a-new-default-model-for-chatgpt/"
L745: }
...
L791: question: "what is the hallucination rate of Claude in 2026",
L792: answer: `<strong>Claude's hallucination rate in 2026 spans from 0% (Claude Opus 4.1 on AA-
High
Browser Session Account Hijack
Browser source reuses an authenticated session to collect identity data and mutate account settings while reporting externally.
dist/server-5WPS3LZD.jsView on unpkg · L252Findings
8 High5 Medium6 Low
HighChild Processdist/server-5WPS3LZD.js
HighShell
HighSame File Env Network Executiondist/bin/mcp-scraper-cli.js
HighBrowser Session Account Hijackdist/server-5WPS3LZD.js
HighCredential Redirect Persistencedist/server-5WPS3LZD.js
HighCommand Output Exfiltrationdist/server-5WPS3LZD.js
HighSandbox Evasion Gated Capabilitydist/bin/mcp-scraper-cli.js
HighSemantic Analysis Limiteddist/chunk-K7Y27ND3.js
MediumNetwork
MediumEnvironment Vars
MediumOversized Source Filedist/bin/api-server.cjs
MediumOversized Cli Entrypointdist/bin/api-server.cjs
Medium
Affected versions and remediation
This report applies to mcp-scraper@0.66.6.
See all versions of mcp-scraper for other recorded verdicts.
- Review the evidence and your use of mcp-scraper@0.66.6 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.