Skip to main content
Glama

webdatatools MCP server

An MCP (Model Context Protocol) server that gives AI agents — Claude Desktop, Cursor, Cline, ChatGPT desktop, or any other MCP client — eleven web-data tools backed by WebDataTools' Apify Actors: web search, article extraction, site crawling, contact/tech-stack detection, company profiling, e-mail/DNS security checks, e-mail validation, package health, and raw Google search.

This server uses your own Apify API token. Every tool call runs an Actor under your Apify account and is billed to your Apify credit, not ours. Nothing is sent anywhere else.

All of the Actors behind these tools are listed at webdatatools.

Requirements

Related MCP server: ToolCenter MCP

Quick start (no install)

npx -y github:paulet4a-commits/webdatatools-mcp-server

The server speaks MCP over stdio. It reads your token from the APIFY_TOKEN environment variable. If APIFY_TOKEN isn't set, the server still starts and lists all eleven tools (so it doesn't look broken in a client's UI) — calling a tool without a token returns a friendly error telling you where to get one.

Claude Desktop

Edit claude_desktop_config.json (Settings → Developer → Edit Config) and add:

{
  "mcpServers": {
    "webdatatools": {
      "command": "npx",
      "args": ["-y", "github:paulet4a-commits/webdatatools-mcp-server"],
      "env": {
        "APIFY_TOKEN": "apify_api_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

Restart Claude Desktop. The eleven tools below appear under the "webdatatools" server.

Cursor

Cursor → Settings → MCP → Add new MCP server, or edit .cursor/mcp.json in your project (or the global ~/.cursor/mcp.json):

{
  "mcpServers": {
    "webdatatools": {
      "command": "npx",
      "args": ["-y", "github:paulet4a-commits/webdatatools-mcp-server"],
      "env": {
        "APIFY_TOKEN": "apify_api_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

Cline (VS Code extension)

Cline → MCP Servers → Configure MCP Servers, which opens cline_mcp_settings.json:

{
  "mcpServers": {
    "webdatatools": {
      "command": "npx",
      "args": ["-y", "github:paulet4a-commits/webdatatools-mcp-server"],
      "env": {
        "APIFY_TOKEN": "apify_api_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
      },
      "disabled": false
    }
  }
}

Any other MCP client that supports a stdio server with an env map works the same way — set command to npx, args to ["-y", "github:paulet4a-commits/webdatatools-mcp-server"], and pass APIFY_TOKEN in env.

Tools

Tool

Actor

What it does

web_search

ai-web-search

Search the web (or read specific URLs) and get clean Markdown of the top results

read_url

article-extractor

Fetch article/blog/news URL(s) and return clean readable Markdown

crawl_site_markdown

website-to-markdown

Crawl a site from a start URL and return one Markdown doc per page (maxPages controls cost)

extract_contacts

contact-extractor

Pull e-mails, phone numbers and social links off a website

detect_tech_stack

tech-stack-detector

Fingerprint CMS, e-commerce, analytics, ad pixels, chat, payments, frameworks, CDN

extract_with_selectors

css-selector-extractor

Extract exact fields from any page with CSS selectors; list pages, pagination, sitemaps (billed per page)

company_profile

company-360

One-call company profile: contacts, tech, DNS/e-mail security, TLS grade, hiring, Wikidata facts, SEO

check_email_security

dns-email-security-checker

SPF/DMARC/DKIM, mail & DNS provider, registrar, domain age, 0-100 score

validate_emails

email-validator

Syntax + disposable/role detection + live MX/A lookup for a list of e-mails

package_health

package-health-checker

npm/PyPI/Crates.io package health: version, downloads, license, GitHub stats

google_search

google-search-scraper

Raw Google SERP data: organic results, People Also Ask, related searches

Each tool has a precise JSON-Schema inputSchema — the calling agent reads the field descriptions to decide when and how to call it. Most tools accept either a singular convenience field (url, domain, email, package, query) or a plural batch array (urls, domains, emails, packages, queries); the plural form wins when both are given.

More WebDataTools MCP servers

Every WebDataTools Actor is also available as a tool, grouped by theme:

Pricing

There is no charge for the MCP server itself. Each tool call runs the underlying Actor on your Apify account and is billed at our Apify Store price for that Actor (pay-per-result or pay-per-event, shown on each Actor's page). Apify's free plan includes $5 of platform credit per month, which comfortably covers light use of these tools. See each Actor's Store page (linked above) for exact pricing, or https://console.apify.com/billing for your own usage and balance.

How results are kept small

Tool results are dataset rows from the Actor run, trimmed for readability:

  • any text field longer than ~4,000 characters is truncated with a note of how many characters were cut

  • the whole response is capped at roughly 100 KB; if it doesn't fit, rows are dropped from the end and the reply says how many of the total rows were included vs. omitted

If you need the full untrimmed data, run the Actor directly from the Apify Console or API.

Configuration reference

Env var

Required

Description

APIFY_TOKEN

Yes (for tool calls)

Your personal Apify API token. Tool discovery works without it; calling a tool without it returns a friendly error.

Per-call timeout is 120 seconds. If an Actor run doesn't finish in time, the tool returns a timeout error suggesting a smaller request (fewer URLs/pages/items).

Local development

npm install
npm test          # vitest, all HTTP mocked — no live network or token needed
npm start          # run the server on stdio (needs APIFY_TOKEN in the environment)

Publishing (not yet published — maintainer decision)

Install straight from GitHub — no npm account or publish step needed:

npx -y github:paulet4a-commits/webdatatools-mcp-server

Publishing to npm would only shorten that to npx -y @webdatatools/mcp-server; the GitHub form above is verified working and is what the config examples use.

Listing on MCP directories

  • Smithery (https://smithery.ai): the repo already includes smithery.yaml. Connect the GitHub repo from the Smithery dashboard ("Deploy Server") — nothing else to configure, it reads smithery.yaml for the stdio start command and the apifyToken config field.

  • Glama (https://glama.ai/mcp/servers): Glama auto-indexes public GitHub repos that ship a valid package.json with an MCP server bin entry and a README with a config example (this repo has both) plus glama.json declaring the maintainer. Submit the repo URL from https://glama.ai/mcp/servers/new once it's pushed to GitHub.

  • mcp.so (https://mcp.so): submit the GitHub repo URL and/or npm package name via their "Submit" form; it scrapes package.json, README.md and the tool list the same way.

None of the above were created or submitted as part of this task — see the final report for what's left to do by hand.

Available Tools

10 tools
check_email_securityB

Check a domain's DNS and e-mail security posture: SPF, DMARC, DKIM selector presence, mail/DNS provider, registrar and domain age, with an overall 0-100 score.

Inputs: domain, domains, checkDkim.

company_profileA

Build a one-stop company profile for a domain: contacts, tech stack, DNS/e-mail security, TLS/security grade, hiring signals, Wikidata company facts (HQ, employees, revenue, founders) and an SEO audit. Turn off sections you don't need to make the run cheaper and faster.

Inputs: domain, domains, includeSeo, includeHiring, +5 more.

crawl_site_markdownA

Crawl a website starting from one URL and return one Markdown document per page (docs sites, blogs, knowledge bases). Billed per page crawled, so keep maxPages as low as the task allows.

Inputs: url (required), maxDepth, maxPages, useSitemap, +2 more.

detect_tech_stackB

Fingerprint a website's technology stack: CMS, e-commerce platform, analytics, ad pixels, e-mail marketing, chat widgets, payment providers, front-end frameworks, cookie consent, CDN/hosting and web server.

Inputs: url, urls, categories, followSubdomains, +1 more.

extract_contactsB

Crawl a website's home/about/contact pages and pull out e-mail addresses, phone numbers and social profile links (LinkedIn, X/Twitter, Instagram, Facebook, YouTube, TikTok, GitHub, Telegram, WhatsApp).

Inputs: url, urls, maxDepth, extractEmails, +4 more.

package_healthA

Check the health of npm, PyPI or Crates.io packages: latest version, publish recency, download stats, license, and (optionally) GitHub stars/forks/open-issues/archived status. Prefix a name with its registry, e.g. pypi:requests.

Inputs: package, packages, enrichGithub, defaultRegistry.

read_urlA

Fetch one or more article/blog/news URLs and return clean, readable article text as Markdown (title, author, date, body). Use this instead of web_search when you already have the exact URL to read.

Inputs: url, urls, outputFormat, includeImages.

validate_emailsA

Validate a list of e-mail addresses: syntax check, disposable/role-account detection and (optionally) a live MX/A DNS lookup to confirm the domain can receive mail. Good for cleaning a lead list before sending.

Inputs: email, emails, checkMx.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.6/5.0

    Scored across 10 tools

    Disambiguation3/5

    web_search and google_search are distinct (page content vs SERP rankings), but web_search can also read URLs directly, overlapping with read_url. company_profile bundles check_email_security, detect_tech_stack, and extract_contacts, so an agent must decide between the aggregate and individual tools.

    Naming Consistency4/5

    All names use snake_case, but the pattern is mixed: most are verb_noun (check_email_security, validate_emails, read_url, extract_contacts, detect_tech_stack, crawl_site_markdown) while several are noun_noun (web_search, google_search, company_profile, package_health). Still readable and consistent in casing.

    Tool Count5/5

    10 tools is well within the 3–15 sweet spot for a web data toolkit, and each covers a distinct capability (search, crawl, extract, validate, profile). No obvious redundancy despite company_profile bundling others.

    Completeness4/5

    Covers search, URL reading, crawling, contact extraction, tech detection, email security/validation, company profiling, package health, and SERP data. Minor gaps exist (e.g., structured data extraction, sitemap parsing, or a dedicated TLS check), but core web-research workflows are fully supported.

    Maintenance

    ActivitySlowing
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      Not graded
      quality
      D
      maintenance
      MCP server giving AI agents real-time web search, page scraping, company intelligence, email discovery, local lead generation, and a persistent knowledge graph. Pay only for what you use, no subscriptions.
      7 npm
      MIT
    • A
      license
      A
      quality
      D
      maintenance
      A comprehensive MCP server providing 15 web tools including search, scraping, screenshots, SEO audits, and DNS/SSL checks through a single installation. It delivers clean, LLM-optimized outputs so AI agents can focus on reasoning rather than parsing raw HTML.
      15
      17 npm
      MIT
    • A
      license
      Not graded
      quality
      C
      maintenance
      MCP tool server that gives any AI agent the ability to search, scrape, and analyze content across the internet.
      42
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      MCP server that gives AI agents reliable web access with self-healing, anti-bot bypass and clean content extraction. It provides 11 tools like fetch, scrape, and search, metered by credit with cost transparency.
      1
      MIT