# B2B Contact MCP Tool — Apollo/Hunter Alt + SMTP Verified (`harvestlab/contact-extractor`) Actor From $0.0015/lead — 6.7× cheaper than leads-finder $1.5/1k. $0.01/contact for 99%+ run success — SMTP-verified emails + libphonenumber phones + 175+ tech stack. RAG-friendly. AI role + primary-contact scoring. Apollo ($99)/Hunter ($34) alt. x402 + Skyfire-ready. EU/Dutch + GDPR. - **URL**: https://apify.com/harvestlab/contact-extractor.md - **Developed by:** [Nick](https://apify.com/harvestlab) (community) - **Categories:** Lead generation, Business, MCP servers - **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, NaN bookmarks - **User rating**: No ratings yet ## Pricing Pay per usage This Actor is paid per platform usage. The Actor is free to use, and you only pay for the Apify platform usage, which gets cheaper the higher subscription plan you have. Learn more: https://docs.apify.com/platform/actors/running/actors-in-store#pay-per-usage ## What's an Apify Actor? Actors are a software tools running on the Apify platform, for all kinds of web data extraction and automation use cases. In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours, and optionally produces a well-defined JSON output, datasets with results, or files in key-value store. In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server. Actors are written with capital "A". ## How to integrate an Actor? If asked about integration, you help developers integrate Actors into their projects. You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready. The best way to integrate Actors is as follows. In JavaScript/TypeScript projects, use official [JavaScript/TypeScript client](https://docs.apify.com/api/client/js.md): ```bash npm install apify-client ``` In Python projects, use official [Python client library](https://docs.apify.com/api/client/python.md): ```bash pip install apify-client ``` In shell scripts, use [Apify CLI](https://docs.apify.com/cli/docs.md): ````bash # MacOS / Linux curl -fsSL https://apify.com/install-cli.sh | bash # Windows irm https://apify.com/install-cli.ps1 | iex ```bash In AI frameworks, you might use the [Apify MCP server](https://docs.apify.com/platform/integrations/mcp.md). If your project is in a different language, use the [REST API](https://docs.apify.com/api/v2.md). For usage examples, see the [API](#api) section below. For more details, see Apify documentation as [Markdown index](https://docs.apify.com/llms.txt) and [Markdown full-text](https://docs.apify.com/llms-full.txt). # README ## B2B Contact Finder — Apollo/Hunter Alt + SMTP Verified > 🤖 **MCP-Ready** — works as a tool for Claude, Cursor, ChatGPT, and other AI agents via the [Apify MCP Server](https://github.com/apify/apify-mcp-server). Typed inputs + structured outputs (dataset_schema) deployed in Cycle 73. > 🧰 **Works with**: LangChain · LangGraph · CrewAI · LlamaIndex · PydanticAI · Mastra AI · Bee AI — call this actor as a tool from any [Apify MCP](https://github.com/apify/apify-mcp-server) or [actor-templates](https://github.com/apify/actor-templates)-backed AI framework. > 💰 **Pay-per-result, not per compute unit** — you know your cost BEFORE pressing Run. No surprise compute charges, no monthly minimums, no SaaS subscription. Compare to Apify's default \$0.20/CU model where cost depends on actor runtime, memory, and data volume. > 🔁 **Apify rental alternative** — Apify is sunsetting the rental pricing model in October 2026. This Actor is already 100% pay-per-event: drop in as a 1:1 replacement for any rental-based competitor without a contract or monthly minimum. > 🍪 **No cookies, no OAuth, no API keys (for the scraping)** — runs on public data only. AI keys are optional and only needed when `enableAiAnalysis: true`. **$0.01 per contact. SMTP-verified mailboxes. CRM-ready in seconds.** Apollo.io starts at **$99/mo (7,200 credits/yr)**, Hunter.io at **$34/mo (5,000 searches)**, ZoomInfo runs **$15k+/yr** with annual contracts. This Actor is **pay-as-you-go** — pay $0.01 per site enriched, $0.02 per definitively-verified mailbox, no seats, no contract, no monthly minimum. Run once for a 50-account ABM list, scale to 50,000 domains for a market map; the unit cost is identical. ### What it does Extract emails with **real SMTP RCPT TO mailbox verification**, libphonenumber-validated phone numbers (E.164 + country + line-type), **16 social media profiles**, physical addresses, and **175+ tech stack signals** from any public website. Give the Actor a list of domains or URLs and get structured, CRM-ready contact data back in seconds. Built for SDRs, AEs, RevOps, recruiters, ABM marketers, and M&A researchers who need fast, accurate website enrichment without an annual seat license. For each domain you provide, the Actor fetches the homepage and up to 20 subpages (auto-prioritizing `/contact`, `/about`, `/imprint`), extracts emails from `mailto:` links, visible text, and JSON-LD structured data, validates phone numbers via libphonenumber, identifies 16 social profiles, parses physical addresses, detects 175+ tech stack signals, and optionally runs a real **async SMTP RCPT TO probe** and AI role classification. All output is structured JSON, immediately importable into HubSpot, Salesforce, Pipedrive, Outreach, Salesloft, Apollo, Google Sheets, Airtable, or any CRM via Apify's native integrations and webhooks. ### Features - **Deep SMTP RCPT TO mailbox probe (v1.7)** — Opt-in `deepEmailVerification: true` layers a real async SMTP RCPT TO handshake on top of the MX tier. Every MX-cleared email gets classified as **`deliverable`** / **`undeliverable`** / **`catchall`** / **`greylisted`** / **`port_blocked`** / **`rate_limited`** / **`error`**, with the remote SMTP response code and message captured verbatim. The probe is reputation-safe: per-MX-host probe history persists across runs in a named key-value store, capped at **10 probes per host per UTC day** with a **10-second cooldown** between probes to the same host. Catch-all detection runs a second RCPT against a random non-existent local-part on the same connection and downgrades the verdict when the server accepts it. Charged $0.02 per **definitive** verdict (deliverable or undeliverable only) — all other verdicts are free. Requires `verifyEmails: true`. - **Email deliverability verification (v1.6)** — Opt-in `verifyEmails: true` runs a real DNS MX lookup on every extracted email's domain (via [dnspython](https://www.dnspython.org/)), detects common-domain typos via Levenshtein distance (`gmial.com` → `gmail.com`, `outllok.com` → `outlook.com`), flags free-inbox hosts (Gmail/Yahoo/Outlook/iCloud/ProtonMail) and disposable/burner mail (Mailinator/10minutemail/...), and tags every entry **HIGH / MEDIUM / LOW / UNKNOWN** deliverability. Emits a new `emails_verified` output field with `mx_valid`, `mx_hosts`, `disposable`, `free_inbox`, `role_based`, `typo_suggestion`, and a `deliverability` tier on every email. Charged $0.01 per MX-cleared email (typo / disposable / no-MX emails are free). - **Phone validation (v1.5)** — Extracts phone numbers in international formats, then runs every candidate through [libphonenumber](https://github.com/google/libphonenumber) (via the `phonenumbers` Python port). Numbers that pass strict carrier-format validation are emitted as `phones_validated` with E.164, ISO country code, and line-type (MOBILE / FIXED_LINE / TOLL_FREE / VOIP / ...). Numbers that parse but fail regional format rules end up in `phones_uncertain` so you can triage them separately. - **Social media profiles (16 platforms)** — Identifies LinkedIn, Twitter/X, Facebook, Instagram, YouTube, GitHub, TikTok, Pinterest, Telegram, Discord, Threads, **WhatsApp**, **Snapchat**, **Vimeo**, **Twitch**, and **Mastodon** profiles linked from the website. (v1.9: +5 platforms — parity with vdrmota's 15+ claim, plus Mastodon.) - **Physical address parsing** — Locates business addresses from JSON-LD structured data, Schema.org markup, and HTML content. - **Contact-page URL capture** — Returns the canonical /contact or /kontakt page URL so you can deep-link prospects straight into the form. - **Tech stack detection (175+ signals)** — Identifies technologies across CMS platforms, JavaScript frameworks, CSS frameworks, analytics tools, chat widgets, CDNs, and e-commerce solutions. - **Smart subpage crawling** — Automatically visits /contact, /about, /imprint, and similar pages where contact information typically lives. - **JSON-LD and Schema.org parsing** — Extracts structured business data embedded in the page source that is invisible to casual browsing. - **Domain input support** — Enter bare domains like "example.com" without needing to add the protocol. HTTPS is added automatically. - **Duplicate handling** — Automatically deduplicates when the same URL appears in both `urls` and `domains` inputs. - **Optional AI contact enrichment** — When enabled (`enableAiAnalysis`), an LLM classifies each email by role (sales, support, hr, legal, executive, general, personal), groups near-duplicate team addresses (`dedup_groups`), flags non-monitored no-reply inboxes (`deliverability_flags`), picks the single best `primary_contact` for B2B outreach, and picks the single best `primary_phone` from the validated set (prefers MOBILE / FIXED_LINE / TOLL_FREE over PREMIUM_RATE / PAGER / UNKNOWN, prefers numbers whose country matches the domain geo). Supports OpenRouter, Anthropic, Google AI, OpenAI, or self-hosted Ollama. #### What makes this different from other Apify contact scrapers - **16 social platforms vs 6** — LinkedIn (with company-profile-over-personal selection), Twitter/X, Facebook, Instagram, YouTube, GitHub, TikTok, Pinterest, Telegram, Discord, Threads, WhatsApp, Snapchat, Vimeo, Twitch, Mastodon. Competitors typically stop at the first six. (v1.9 expanded 11 → 16.) - **Carrier-verified phone split** — `phones_validated` gives you E.164 + country + line-type per number via libphonenumber. Most Apify contact scrapers ship a single raw `phones` list and leave cleanup to you. - **Real-time MX + SMTP deliverability** — every email can be checked against real DNS MX records AND a real async SMTP RCPT TO probe against the primary MX host. Tagged HIGH / MEDIUM / LOW / UNKNOWN at the MX layer, then `deliverable` / `undeliverable` / `catchall` / `greylisted` / `port_blocked` at the mailbox layer. Typo suggestions, disposable-host detection, reputation-safe per-MX-host probe rate limits (10/day, 10s cooldown, persisted across runs). **No other Apify contact scraper runs a live SMTP handshake on scraped emails.** - **175+ tech-stack detectors** on the same fetch — one run gets you contacts + CMS + analytics + chat + CRM + payment stacks. Competitors either do contacts *or* tech but rarely both. - **Integrated AI enrichment** — role classification, dedup, primary-contact + primary-phone pick, deliverability flags. No other contact-extractor on the Store bundles an LLM enrichment layer. - **5-provider AI choice** — OpenRouter (default, cheapest), Anthropic, Google AI, OpenAI, or self-hosted Ollama for zero-egress setups. #### Head-to-head: vs the dominant Apify contact scraper This Actor competes directly against `vdrmota/contact-info-scraper` (the most-installed contact actor on the Apify Store). Where each wins: | Feature | This Actor (harvestlab) | vdrmota/contact-info-scraper | |---|---|---| | **SMTP RCPT TO mailbox verification** | ✅ 4-tier (deliverable/undeliverable/catchall/greylisted) on every email | ❌ third-party "lead enrichment" only | | **libphonenumber phone validation** | ✅ E.164 + ISO country + line-type (MOBILE/FIXED_LINE/TOLL_FREE/...) | ❌ raw phone list, no validation | | **Tech stack detection** | ✅ 175+ signals on the same fetch (CMS, frameworks, CDN, analytics, payment) | ❌ not offered | | **AI email-role classifier** | ✅ built-in (sales / support / hr / legal / executive / general / personal) | ❌ not offered | | **Multi-LLM provider choice** | ✅ 5 providers (OpenRouter/Anthropic/Google/OpenAI/Ollama) | ❌ no LLM layer | | **Social platforms** | **16** (LinkedIn, X, FB, IG, YT, GitHub, TikTok, Pinterest, Telegram, Discord, Threads, WhatsApp, Snapchat, Vimeo, Twitch, Mastodon) | 15+ | | **Per-result price** | $0.01 / verified contact | $0.00105 / page | | **Rating on Apify Store** | New / unrated | 3.4/5 across 77 reviews | **Why pay 10× more per result?** Because `$0.00105 × 1,000 unvalidated raw pages` is not the same product as `$0.01 × 1,000 SMTP-verified, libphonenumber-validated, AI-role-classified contacts`. We deliver outreach-ready contacts; competitors deliver pages-with-text. Your CRM doesn't want pages — it wants verified mailboxes that won't bounce. **If your priority is raw-page volume and you'll do verification yourself downstream**, vdrmota's actor is cheaper. **If your priority is contacts that don't bounce on the first send**, this Actor's $0.02 SMTP charge is the cheapest mailbox-verification primitive on the Apify Store. ### Use Cases #### SDR / AE Outbound Prospecting — replace Apollo & Hunter at the unit-cost level Enrich prospect lists with **SMTP-verified emails** before outreach. A 1,000-domain prospect list costs roughly **$10 + $20 deep-verify = $30** here vs **~$99/mo flat** on Apollo Starter (and Apollo's data is bought, not freshly extracted from the site). Upload target company domains and get emails, libphonenumber-validated phones, LinkedIn / 16 socials, addresses, and 175+ tech-stack signals — drop straight into HubSpot, Salesforce, Pipedrive, Outreach, or Salesloft via Apify's native integrations. AI role-classifier flags `info@` / `support@` / `noreply@` so you only blast decision-makers. #### Account-Based Marketing (ABM) Lists — match-rate by domain, not name + company guess Build enriched ABM target lists from a domain seed (e.g. "all SaaS companies running Segment + Marketo on Shopify"). Tech stack detection (175+ signals) lets you filter by **CMS, analytics, CRM, chat widget, payment processor, CDN, search platform** — Hunter and Apollo can't do this; ZoomInfo charges five figures/yr to come close. Combine with `enableAiAnalysis` for primary-contact selection per account. #### Lead-Gen Agencies & RevOps — pay-per-event beats per-seat Run hundreds of client enrichment jobs without buying $99-$1,200/mo seats per analyst. **Pay only for sites that returned data**: failed fetches and DNS errors are free. A 10,000-domain quarterly enrichment run costs ~**$100** instead of a Hunter Business plan ($104/mo, 10k searches/mo) — and you keep the raw HTML provenance for every record. #### Recruiter Sourcing & Talent Intel — find decision-makers via tech-stack match Identify hiring managers at companies running specific stacks (e.g. "all React + Next.js + GraphQL shops in NYC"). Tech-stack detection reveals what languages and frameworks a company actually uses *today*, which is more current than LinkedIn skill-tags. Pair with social profile capture (GitHub, LinkedIn) to surface candidate-targeting signals before you pay LinkedIn Recruiter's $11k/yr seat. #### M&A Due Diligence & Investor Research — rapid target screening Run a watchlist of acquisition targets through the Actor every quarter. Output captures **company description, address, contact emails, phone numbers, social footprint, and complete tech stack** — letting M&A analysts and PE / VC associates screen 500-2,000 companies in an afternoon for tech debt, vendor lock-in, growth signals (e.g. "still on jQuery + Bootstrap", "migrated from Magento to Shopify Plus"), and digital maturity. #### CRM Data Hygiene & Re-enrichment — close the 30% empty-field gap Most B2B CRMs have **~30% missing or stale fields** on free-tier records. Upload domains for any CRM segment with empty `phone`, `email`, `linkedin_url`, or `tech_stack` fields and bulk-update with fresh, SMTP-verified data. The dual `phones` + `phones_validated` shape lets you pick raw-extract vs carrier-verified per workflow. AI dedup groups `info@` / `hello@` / `contact@` into a single primary contact per account. #### Competitive Intelligence & Tech-Stack Tracking Schedule regular scans on competitor domains. Detect when a target migrates CMS, swaps analytics, adds a CDP, or rolls out a new payment processor. The `tech_stack` array surfaces 175+ technologies — competitors that ship "tech detection" usually run 30-50 signals; this Actor catches CDP layers (Segment, Pendo, PostHog), product-analytics, customer-data infra, search platforms (Algolia, Typesense), build tools (Vite, Webpack), and feature-flag systems (LaunchDarkly). #### Market Research & Industry Datasets Build structured datasets of companies in a niche by feeding a domain list and exporting JSON / CSV / Excel. Combine tech stack adoption with contact data to segment markets by technology fingerprint, company digital maturity, and likely deal size — perfect for analyst reports, investor decks, and trend studies. ### Input | Parameter | Type | Default | Description | |-----------|------|---------|-------------| | `urls` | array | `[]` | List of full website URLs (e.g., `https://example.com`). | | `domains` | array | `[]` | List of bare domains (e.g., `example.com`, `stripe.com`). HTTPS is added automatically. | | `maxPagesPerSite` | integer | `5` | Number of pages to crawl per site (1–20). More pages means more contact data found but longer runtime. | | `includeSubpages` | boolean | `true` | Automatically crawl /contact, /about, /imprint, and similar pages. Highly recommended — many sites only list contact info on subpages. | | `detectTechStack` | boolean | `true` | Detect CMS, frameworks, analytics, and 175+ other technologies. Set to `false` for faster runs when you only need contact data. | | `verifyEmails` | boolean | `false` | Run real DNS MX lookups on every extracted email, detect typos / disposable / free-inbox domains, and tag each email HIGH / MEDIUM / LOW / UNKNOWN deliverability. Adds the `emails_verified` output field. Cost: $0.01 per MX-cleared email. | | `deepEmailVerification` | boolean | `false` | Opt-in: real async SMTP RCPT TO probe against the primary MX host for every MX-cleared email. Classifies each as `deliverable` / `undeliverable` / `catchall` / `greylisted` / `port_blocked` / `rate_limited`. Reputation-safe: 10 probes/host/day cap + 10s cooldown, persisted across runs. Charges $0.02 per definitive verdict only. Requires `verifyEmails: true`. | | `enableAiAnalysis` | boolean | `false` | Enable AI-powered contact enrichment — email role classification, team-address deduplication, primary-contact selection, and deliverability flags. Requires an API key for your chosen LLM provider. | | `llmProvider` | string | `openrouter` | AI provider when enrichment is enabled: `openrouter`, `anthropic`, `google`, `openai`, or `ollama`. | | `openrouterApiKey` / `anthropicApiKey` / `googleApiKey` / `openaiApiKey` | string | — | API key for the chosen provider. Ollama (self-hosted) uses `ollamaBaseUrl` instead. | | `proxyConfiguration` | object | — | Apify proxy settings for accessing geo-restricted or bot-protected sites. | You can provide input via `urls`, `domains`, or both. Duplicates across both inputs are removed automatically. #### Pricing This Actor uses **pay-per-event (PPE)** pricing. | Event | Price | Description | |-------|-------|-------------| | Contact extracted | $0.01 | Per website successfully processed and contacts extracted | | Verified phone extracted | $0.02 | Charged once per libphonenumber-validated phone number (E.164 + country + line-type). Only fires on numbers that pass strict carrier-format validation. | | Verified email checked | $0.01 | Charged once per email whose domain passes real MX-record verification (HIGH / MEDIUM tier). Typo / disposable / no-MX emails (LOW tier) and UNKNOWN results are free. Only fires when `verifyEmails` is on. | | Verified email deep-checked | $0.02 | Charged once per email that receives a **definitive** SMTP RCPT verdict (`deliverable` or `undeliverable`). `catchall`, `greylisted`, `rate_limited`, `port_blocked`, and `error` verdicts are free. Only fires when `deepEmailVerification` is on. | | AI analysis completed | $0.05 | Per website when `enableAiAnalysis` is on (role classification, email dedup, primary-contact + primary-phone selection, deliverability flags). | Failed requests (unreachable sites, DNS errors) are not charged. | Scenario | Websites | Estimated Cost | |----------|----------|----------------| | Quick test batch | 10 | ~$0.10 | | Sales prospect list | 100 | ~$1.00 | | Market research dataset | 500 | ~$5.00 | | Large-scale enrichment | 2,000 | ~$20.00 | Plus Apify platform compute costs. A typical batch of 100 websites completes in 3–8 minutes depending on site responsiveness and pages crawled per site. ### Use with AI agents Contact-extractor returns structured contact dicts (`emails`, `email_roles`, `primary_contact`, `phones_validated`, `tech_stack`, `social_profiles`, `verified` flags) — the exact shape an outbound-outreach LLM agent needs to plan, qualify, and personalize. Pair it with the new Apify [`js-langchain`](https://github.com/apify/actor-templates) and [`js-langgraph-agent`](https://github.com/apify/actor-templates) templates: agents can iterate cheaply at **$0.0015/lead** via the lead-discovery tier, then verify the best targets at **$0.01/contact** with the full SMTP-verified pipeline. #### LangChain — wrap as a Tool ```python from apify_client import ApifyClient from langchain.tools import Tool client = ApifyClient("YOUR_APIFY_TOKEN") def find_b2b_contacts(start_urls: list[str]) -> list[dict]: run = client.actor("harvestlab/contact-extractor").call( run_input={"urls": start_urls, "verifyEmails": True, "enableAiAnalysis": True} ) return list(client.dataset(run["defaultDatasetId"]).iterate_items()) contact_tool = Tool( name="find_b2b_contacts", description="Extract SMTP-verified B2B contacts (email, role, phone, tech stack) from website URLs.", func=find_b2b_contacts, ) ## agent.invoke({"input": "Find decision-maker contacts at stripe.com and shopify.com"}) contact_tool.invoke({"startUrls": ["https://stripe.com", "https://shopify.com"]}) ```` #### LangGraph — node in a sales-outreach StateGraph ```python from typing import TypedDict from apify_client import ApifyClient from langgraph.graph import StateGraph, END class OutreachState(TypedDict): domain: str contacts: list[dict] client = ApifyClient("YOUR_APIFY_TOKEN") def enrich_contacts(state: OutreachState) -> OutreachState: run = client.actor("harvestlab/contact-extractor").call( run_input={"domains": [state["domain"]], "verifyEmails": True, "enableAiAnalysis": True} ) items = list(client.dataset(run["defaultDatasetId"]).iterate_items()) verified = [e for item in items for e in item.get("emails_verified", []) if e.get("mx_valid")] return {**state, "contacts": verified} graph = StateGraph(OutreachState) graph.add_node("enrich", enrich_contacts) graph.set_entry_point("enrich") graph.add_edge("enrich", END) ## graph.compile().invoke({"domain": "stripe.com", "contacts": []}) ``` See Apify's [`actor-templates`](https://github.com/apify/actor-templates) repo for the full `js-langchain` and `js-langgraph-agent` starters — drop this actor in as a tool and the agent handles ICP filtering, role-aware drafting, and follow-up sequencing. ### Output Each website produces a structured JSON object: ```json { "url": "https://example.com", "domain": "example.com", "company_name": "Example Inc", "company_description": "Example Inc is a technology company that helps businesses build better software solutions and streamline their operations.", "emails": ["info@example.com", "sales@example.com"], "emails_verified": [ { "email": "info@example.com", "domain": "example.com", "mx_valid": true, "mx_hosts": ["aspmx.l.google.com", "alt1.aspmx.l.google.com"], "catchall": false, "disposable": false, "free_inbox": false, "role_based": true, "typo_suggestion": null, "deliverability": "MEDIUM", "deep_probe": true, "probe_result": "deliverable", "smtp_probe": { "result": "deliverable", "response_code": 250, "response_text": "2.1.5 OK", "mx_host": "aspmx.l.google.com", "probed_at": "2026-04-23T03:15:22+00:00", "is_catchall": false } }, { "email": "sales@example.com", "domain": "example.com", "mx_valid": true, "mx_hosts": ["aspmx.l.google.com", "alt1.aspmx.l.google.com"], "catchall": null, "disposable": false, "free_inbox": false, "role_based": true, "typo_suggestion": null, "deliverability": "MEDIUM", "deep_probe": true, "probe_result": "rate_limited", "smtp_probe": { "result": "rate_limited", "response_code": null, "response_text": "daily_cap_reached:10/10", "mx_host": "aspmx.l.google.com", "probed_at": "2026-04-23T03:15:22+00:00", "is_catchall": null } } ], "phones": ["+1 888 926 2289", "555-1234"], "phones_validated": [ { "raw": "+1 888 926 2289", "e164": "+18889262289", "country": "US", "type": "TOLL_FREE", "valid": true, "possible": true } ], "phones_uncertain": ["555-1234"], "social_profiles": { "linkedin": "https://linkedin.com/company/example", "twitter": "https://twitter.com/example", "facebook": null, "instagram": null, "youtube": null, "github": "https://github.com/example", "tiktok": "https://tiktok.com/@example", "pinterest": null, "telegram": "https://t.me/examplecompany", "discord": null, "threads": "https://threads.net/@example" }, "address": "123 Main St, City, State, 12345", "contact_page_url": "https://example.com/contact", "tech_stack": ["WordPress", "PHP", "nginx", "Google Analytics", "jQuery"], "pages_crawled": 3, "scraped_at": "2026-04-10T12:00:00Z" } ``` **Which phone field should you use?** For carrier-verified dialing, cold-call campaigns, or CRM import, use **`phones_validated`** — every entry carries a canonical E.164 string, ISO-3166-1 country code, and line-type so your dialer knows whether it's mobile, toll-free, or a PBX fixed line. The legacy **`phones`** list (all raw extracts) and **`phones_uncertain`** (parsed but not carrier-format verified) are kept for backward compatibility and triage workflows. Null values indicate that the Actor searched for the data but did not find it on the crawled pages. Output is available as JSON, CSV, or Excel. Use Apify integrations to push results directly to Google Sheets, HubSpot, Salesforce, Airtable, or any webhook endpoint. When `enableAiAnalysis` is `true`, each output item additionally includes: - `email_roles` — per-email classification (sales, support, hr, legal, executive, general, personal) - `primary_contact` — the single best email for B2B outreach - `primary_phone` — the single best validated phone for B2B outreach (picked from `phones_validated`, prefers mobile / fixed-line / toll-free in the domain's country over premium-rate / pager / unknown) - `dedup_groups` — near-duplicate team addresses grouped together (e.g. `info@` and `hello@` pointing to the same inbox) - `deliverability_flags` — warnings for no-reply and non-monitored addresses #### Technologies Detected The Actor recognizes 175+ technologies across these categories: - **CMS and Platforms** — WordPress, Shopify, Wix, Squarespace, Drupal, Joomla, Ghost, Webflow, HubSpot - **JavaScript Frameworks** — React, Next.js, Vue.js, Nuxt.js, Angular, Svelte, jQuery - **CSS Frameworks** — Bootstrap, Tailwind CSS, Foundation, Bulma, Materialize - **Analytics and Marketing** — Google Analytics, Google Tag Manager, Facebook Pixel, Hotjar, Mixpanel, Segment, Plausible, Matomo, Microsoft Clarity, Amplitude, LinkedIn Insight Tag - **Product Analytics** — FullStory, LogRocket, PostHog, Pendo, Datadog RUM, Sentry, LaunchDarkly - **Email Marketing** — Mailchimp, SendGrid, Klaviyo, ActiveCampaign, Marketo, Pardot, ConvertKit - **CRM and Sales** — Salesforce, HubSpot, Pipedrive - **Chat and Support** — Intercom, Drift, Zendesk, Crisp, LiveChat, Tawk.to, Freshdesk - **Authentication** — Auth0, Okta, Firebase, Supabase - **Search** — Algolia, Elasticsearch, Typesense - **Infrastructure** — Cloudflare, Fastly, CloudFront, Vercel, Netlify, nginx, Apache, Akamai - **E-commerce** — WooCommerce, Magento, PrestaShop, BigCommerce, Stripe, PayPal - **Video and Media** — YouTube Embed, Vimeo, Wistia, Loom - **Scheduling** — Calendly, Acuity Scheduling - **Social Proof** — Trustpilot Widget, G2, Yotpo - **Build Tools** — Webpack, Vite, GraphQL, WebSocket ### Quick Start The simplest possible input — a list of domains: ```json { "domains": ["stripe.com", "shopify.com", "hubspot.com"] } ``` Or full URLs with custom crawl depth: ```json { "urls": ["https://example.com", "https://anothersite.org"], "maxPagesPerSite": 10, "includeSubpages": true, "detectTechStack": true } ``` Enable SMTP-verified emails for CRM import: ```json { "domains": ["stripe.com", "shopify.com"], "verifyEmails": true, "deepEmailVerification": true } ``` Enable AI role classification and primary-contact selection: ```json { "domains": ["stripe.com", "shopify.com"], "verifyEmails": true, "enableAiAnalysis": true, "llmProvider": "openrouter", "openrouterApiKey": "your-key-here" } ``` #### MCP Quickstart — call this actor from Claude / Cursor / ChatGPT Install the Apify MCP server in your AI agent of choice: ```bash ## Claude Code claude mcp add apify -- npx -y @apify/actors-mcp-server --token YOUR_APIFY_TOKEN ## Claude Desktop / Cursor (add to mcp.json): {"mcpServers":{"apify":{"command":"npx","args":["-y","@apify/actors-mcp-server","--token","YOUR_APIFY_TOKEN"]}}} ``` Then prompt the agent: > "Use the harvestlab/contact-extractor actor on Apify to extract SMTP-verified contact emails for 50 SaaS companies in \[list of domains] with primary-contact scoring and AI email-role classification. Push the results back as JSON." The agent will discover the actor's `dataset_schema.json`, generate the right input, run it, and pipe the typed output back into your conversation. ### Troubleshooting **SMTP verification returns "greylisted" for most emails** Greylisting is a deliberate delay tactic — mail servers temporarily reject unknown senders and accept on retry. The actor does not retry SMTP probes (that would be too slow). `greylisted` emails are real but the server wasn't conclusive; treat them as "probably valid, low confidence". Retry the same URL in 10–15 minutes for a more definitive result. **Many emails flagged as "catchall" or "undeliverable"** Catchall domains accept all email addresses regardless of whether the mailbox exists (common with corporate domains using wildcard MX routing). These cannot be verified via SMTP RCPT — treat as "valid domain, unknown mailbox". Filter with `deliverabilityFilter: "deliverable_only"` to exclude them. **No tech stack signals returned** Tech detection requires loading the page's HTML and JS assets. Pages behind login walls, heavy JavaScript SPAs, or sites that block headless fetching will return fewer tech signals. The actor detects 175+ technologies from publicly visible assets only. **AI role classification returns wrong roles** The AI role classifier uses the email prefix and page context to infer roles (sales/marketing/engineering/executive/support). It cannot access LinkedIn or internal org data. For ambiguous prefixes (`info@`, `hello@`, `contact@`), it classifies as "unknown" — these are typically shared inboxes. **Rate limited or IP blocked mid-crawl** The actor implements reputation-safe probe rate limits (max 10 SMTP probes per domain per UTC day, 10s cooldown). If a domain blocks the probe IP, the actor logs `smtp_blocked: true` and falls back to MX-only verification. Use a RESIDENTIAL proxy to rotate IPs between domains. **No emails found for a domain** The site may not publish contact emails publicly. Try increasing `maxPagesPerSite` to crawl more pages, or check if the site uses a contact form instead of publishing email addresses. The actor only extracts emails that are publicly visible — it does not guess or generate addresses. **How many pages should I crawl per website?** The default of 5 pages works well for most sites. The Actor prioritizes /contact, /about, and /imprint pages. Set `maxPagesPerSite` to 1 if you only need tech stack data and want the fastest possible run. Increase to 10–20 for large corporate sites where contact details may be spread across many sections. **Can I input just domain names without https://?** Yes. Use the `domains` field with bare domain names like "stripe.com" or "hubspot.com". The Actor adds HTTPS automatically. You can also mix `urls` (full URLs) and `domains` (bare domains) in the same run. **How does email deliverability verification work?** When you set `verifyEmails: true`, every extracted email runs through a four-step pipeline: 1. **Typo check** — The email's domain is compared (bounded Levenshtein distance 1–2) against a dictionary of the 25 most-common consumer and business mail hosts. Near-misses like `gmial.com`, `outllok.com`, `hotmial.com`, `iclould.com` are flagged with a `typo_suggestion` and tagged **LOW** deliverability. 2. **Disposable / burner check** — Known disposable hosts (Mailinator, 10MinuteMail, GuerrillaMail, YOPmail) are hard-flagged **LOW** regardless of MX state. 3. **Real MX lookup** — The Actor runs an actual DNS MX query against the domain. Domains that have no MX record cannot receive mail and are tagged **LOW**. 4. **Tiering** — Emails at free inboxes (Gmail / Yahoo / Outlook / iCloud / ProtonMail) OR with role prefixes (`info@`, `hello@`, `support@`, `sales@`) land in **MEDIUM**. Named addresses at non-free company domains with resolved MX land in **HIGH**. **How does the deep SMTP probe work?** When `verifyEmails` and `deepEmailVerification` are both on, every MX-cleared email gets a real async SMTP RCPT TO handshake: 1. `EHLO` with a neutral sender domain, `MAIL FROM:<>` (null sender), and `RCPT TO:`. No mail is delivered. 2. Response code classification: `250/251` → `deliverable`, `550/551/553/554` → `undeliverable`, `421/450/451/452` → `greylisted`. 3. Catch-all test: a second RCPT against a random non-existent local-part on the same connection — if the server accepts it too, the verdict downgrades to `catchall`. 4. `QUIT` — connection closes without sending mail. Reputation-safety: the probe history persists across runs in a named Apify key-value store, hard-capped at **10 probes per MX host per UTC day** with a **10-second cooldown**. **How does phone validation work?** Every phone candidate is passed through [libphonenumber](https://github.com/google/libphonenumber) — the same library Google Android uses. Numbers that clear both `is_possible_number` and `is_valid_number` land in `phones_validated` with E.164, country code, and line-type. Numbers that parse but fail strict validation land in `phones_uncertain`. This is the same carrier-verified output shape that competitor services charge $8 per 1,000 sites for — here it's baked into the base $0.01 site price. **What if a website blocks the scraper?** Some sites block datacenter IP addresses. Enable Apify proxy configuration with residential proxies to access these sites. Sites behind login walls or with aggressive bot protection may still be inaccessible. Failed requests are not charged. #### Dutch B2B Intelligence Suite Cross-portfolio: this Actor is the contact-enrichment layer of the **Dutch B2B Intelligence Suite** — combine with these niche scrapers: - **[funda-scraper](https://apify.com/harvestlab/funda-scraper)** — Dutch real estate listings, sold prices, rental yield analysis - **[kvk-scraper](https://apify.com/harvestlab/kvk-scraper)** — Dutch KvK Chamber of Commerce business registry - **[marktplaats-scraper](https://apify.com/harvestlab/marktplaats-scraper)** — NL classifieds + 2dehands.be with seller info - **[bol-com-scraper](https://apify.com/harvestlab/bol-com-scraper)** — NL/BE retail products, prices, watch-mode drop alerts - **[gov-procurement-scraper](https://apify.com/harvestlab/gov-procurement-scraper)** — NL Tenders + EU TED + UK + US procurement - **[contact-extractor](https://apify.com/harvestlab/contact-extractor)** — SMTP-verified emails for any `.nl` business website *A Dutch-niche cross-portfolio bundle: 5 specialized scrapers + 1 contact-enrichment layer that no Apify-maintained actor competes with. Together: company registry signal → real-estate footprint → classifieds activity → e-commerce SKU intel → public-sector contracts → SMTP-verified outbound contacts.* ### Legal and Compliance This actor scrapes publicly available data. By using this actor, you agree to the following: - **Your responsibility**: You are solely responsible for ensuring your use complies with all applicable laws, regulations, and the target website's terms of service. This includes but is not limited to **GDPR** (EU), **CCPA / CPRA** (California), **CAN-SPAM Act** (US), **CASL** (Canada), **PECR** (UK), **LGPD** (Brazil), and other data protection / anti-spam laws in your jurisdiction. - **No legal advice**: This actor does not constitute legal advice. Consult a qualified attorney if you have questions about the legality of your specific use case. - **Intended use**: This actor is designed for legitimate business purposes such as market research, competitive analysis, and B2B lead generation using publicly accessible data. - **Data handling**: You are responsible for how you store, process, and share any data collected. Ensure you have a lawful basis (e.g. **legitimate interest** under GDPR Art. 6(1)(f), or pre-existing business relationship under CASL) for processing any personal data under applicable privacy laws. - **CAN-SPAM compliance** (US): Any commercial email you send using contacts from this actor must include a clear, conspicuous opt-out mechanism, a valid physical postal address, accurate "From" / "Reply-To" headers, non-deceptive subject lines, and you must honor unsubscribe requests within 10 business days. Penalties are up to **$53,088 per violation** under the FTC's enforcement schedule. - **GDPR compliance** (EU/UK): Even publicly available personal data is subject to GDPR. You must have a lawful basis (typically **legitimate interest with documented LIA**), respect Art. 14 transparency requirements (notify the data subject within 30 days when collecting from a third-party source), honor erasure / objection requests promptly, maintain records of processing under Art. 30, and not transfer EU data outside the EU/EEA without an SCC, adequacy decision, or other approved mechanism. - **Anti-spam / outreach**: Do **NOT** use this tool for unsolicited bulk messaging, spam, scraped-list email blasts, or list-building for resale. Use for permission-tested, targeted, B2B-relevant outreach only. - **Rate limiting**: This actor implements polite crawling practices including request delays and retry backoff to minimize impact on target servers. - **No warranty**: This actor is provided "as is" without warranty. Data accuracy depends on the target website's content and structure. - **Personal data minimization**: Implement data retention policies (typically 6-12 months for cold outreach contacts), encrypt PII at rest and in transit, restrict internal access on a need-to-know basis, and honor opt-out / Do-Not-Contact requests across your entire system, not just the channel that received the request. - **SMTP probing**: The SMTP RCPT TO probe uses reputation-safe rate limiting (10 probes/host/day, 10s cooldown) and never delivers mail. The null sender (`MAIL FROM:<>`) is the standard non-delivery bounce address used by mailer-daemons. Nevertheless, some mail server operators may log or flag probes — use responsibly. The probe never sends mail content, never authenticates, and runs only on domains you've explicitly listed in `urls` or `domains`. #### Related Actors - **[Indeed Scraper](https://apify.com/harvestlab/indeed-scraper)** — Pipe Indeed job listings into a recruiter sourcing pipeline: scrape hiring companies, then run Contact Extractor on the listing's company URL to get hiring-manager emails and phones for outbound recruiting. - **[News Monitor](https://apify.com/harvestlab/news-monitor)** — Build account intelligence on prospects you've extracted contacts for. Track funding rounds, leadership changes, and product news so SDRs can time outreach around real-world events instead of cold-pitching. - **[Google Search Scraper](https://apify.com/harvestlab/google-search-scraper)** — Discover target domains via SEO research before extracting contacts. Scrape SERPs for "best \[category] software" or competitor intent queries, then feed the top-ranking domains into Contact Extractor for ICP-matched lead lists. # Actor input Schema ## `urls` (type: `array`): List of full website URLs to analyze (e.g. https://example.com). You can provide either URLs or domains. ## `startUrls` (type: `array`): Alternative field name for Website URLs, provided for compatibility with Apify's common 'startUrls' convention (e.g. https://example.com). The canonical field is `urls` above — prefer that. If both are supplied, they are merged. ## `url` (type: `string`): CLI alias for a single website URL. Hidden from Console form. ## `website` (type: `string`): CLI alias for a single website URL. Hidden from Console form. ## `domains` (type: `array`): List of domains without protocol (e.g. example.com). HTTPS will be added automatically. Alternative to providing full URLs — use this OR Website URLs above. ## `maxPagesPerSite` (type: `integer`): Maximum number of pages to crawl per website. Higher values find more contacts but take longer. ## `includeSubpages` (type: `boolean`): Automatically crawl /contact, /about, /imprint, and similar pages to find additional contact information. ## `detectTechStack` (type: `boolean`): Analyze the website to detect technologies used (CMS, frameworks, analytics, etc.). ## `verifyEmails` (type: `boolean`): Look up real MX records for every extracted email, detect common-domain typos (e.g. gmial.com -> gmail.com), flag free/disposable inboxes, and tag each email HIGH / MEDIUM / LOW / UNKNOWN deliverability. Adds a new `emails_verified` field to every output item. Cost: $0.01 per email whose domain passes MX resolution (uncertain / disposable / no-MX emails are free). ## `deepEmailVerification` (type: `boolean`): Opt-in: run an async SMTP RCPT TO probe against the primary MX host for every MX-cleared email. Classifies each email as deliverable / undeliverable / catchall / greylisted / port\_blocked. Reputation-safe: per-MX-host probe history persists across runs (named KV store), capped at 10 probes/host/day with 10s cooldown between probes to the same host. Port 25 is frequently blocked on Apify datacenter egress; when it is, every entry returns `port_blocked` (free). Cost: $0.02 per email that receives a definitive verdict (deliverable or undeliverable) — catch-all, greylisted, and non-verdict results are free. Requires `verifyEmails: true`. ## `enableAiAnalysis` (type: `boolean`): Use an LLM to classify each email by role (sales/support/hr/legal/executive/general/personal), group near-duplicate team addresses, flag non-monitored no-reply inboxes, and pick the single best primary contact for B2B outreach. When `verifyEmails` is also enabled, the AI picks the primary contact only from HIGH/MEDIUM deliverability emails. Requires an API key for your chosen LLM provider. Cost: $0.05 per website analyzed. ## `llmProvider` (type: `string`): AI backend for email role classification, dedup, and primary-contact selection. 'OpenRouter' (default) is cheapest — Gemini Flash via OpenRouter is ~$0.001 per website. 'Anthropic' = Claude, 'Google AI' = Gemini direct, 'OpenAI' = GPT-4o mini, 'Ollama' = self-hosted (no API cost). Each provider needs its own API key field below. ## `llmModel` (type: `string`): Specific model to use. Leave empty for the provider default (google/gemini-2.0-flash-001 for OpenRouter, claude-sonnet-4-20250514 for Anthropic, gemini-2.0-flash for Google AI, gpt-4o-mini for OpenAI, llama3.1 for Ollama). ## `openrouterApiKey` (type: `string`): Your OpenRouter API key. Get one at openrouter.ai/keys ## `anthropicApiKey` (type: `string`): Your Anthropic API key. Get one at console.anthropic.com ## `googleApiKey` (type: `string`): API key for Google AI (Gemini). Get one at aistudio.google.com/app/apikey ## `openaiApiKey` (type: `string`): API key from platform.openai.com (required if using OpenAI provider) ## `ollamaBaseUrl` (type: `string`): Base URL for Ollama API. Default: http://localhost:11434 ## `proxyConfiguration` (type: `object`): Proxy settings. Recommended for reliable results — many websites block requests from datacenter IPs. ## Actor input object example ```json { "urls": [ "https://www.hubspot.com", "https://www.notion.so" ], "domains": [ "hubspot.com", "notion.so" ], "maxPagesPerSite": 5, "includeSubpages": true, "detectTechStack": true, "verifyEmails": false, "deepEmailVerification": false, "enableAiAnalysis": false, "llmProvider": "openrouter", "ollamaBaseUrl": "http://localhost:11434", "proxyConfiguration": { "useApifyProxy": true } } ``` # Actor output Schema ## `datasetOutput` (type: `string`): Dataset containing scraped B2B contact records: SMTP-verified emails, libphonenumber-validated phones, social profiles, tech stack, and AI email-role classifications. # API You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup. ## JavaScript example ```javascript import { ApifyClient } from 'apify-client'; // Initialize the ApifyClient with your Apify API token // Replace the '' with your token const client = new ApifyClient({ token: '', }); // Prepare Actor input const input = { "urls": [ "https://www.hubspot.com", "https://www.notion.so" ], "domains": [ "hubspot.com", "notion.so" ], "ollamaBaseUrl": "http://localhost:11434", "proxyConfiguration": { "useApifyProxy": true } }; // Run the Actor and wait for it to finish const run = await client.actor("harvestlab/contact-extractor").call(input); // Fetch and print Actor results from the run's dataset (if any) console.log('Results from dataset'); console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`); const { items } = await client.dataset(run.defaultDatasetId).listItems(); items.forEach((item) => { console.dir(item); }); // 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs ``` ## Python example ```python from apify_client import ApifyClient # Initialize the ApifyClient with your Apify API token # Replace '' with your token. client = ApifyClient("") # Prepare the Actor input run_input = { "urls": [ "https://www.hubspot.com", "https://www.notion.so", ], "domains": [ "hubspot.com", "notion.so", ], "ollamaBaseUrl": "http://localhost:11434", "proxyConfiguration": { "useApifyProxy": True }, } # Run the Actor and wait for it to finish run = client.actor("harvestlab/contact-extractor").call(run_input=run_input) # Fetch and print Actor results from the run's dataset (if there are any) print("💾 Check your data here: https://console.apify.com/storage/datasets/" + run["defaultDatasetId"]) for item in client.dataset(run["defaultDatasetId"]).iterate_items(): print(item) # 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start ``` ## CLI example ```bash echo '{ "urls": [ "https://www.hubspot.com", "https://www.notion.so" ], "domains": [ "hubspot.com", "notion.so" ], "ollamaBaseUrl": "http://localhost:11434", "proxyConfiguration": { "useApifyProxy": true } }' | apify call harvestlab/contact-extractor --silent --output-dataset ``` ## MCP server setup ```json { "mcpServers": { "apify": { "command": "npx", "args": [ "mcp-remote", "https://mcp.apify.com/?tools=harvestlab/contact-extractor", "--header", "Authorization: Bearer " ] } } } ``` ## OpenAPI specification ```json { "openapi": "3.0.1", "info": { "title": "B2B Contact MCP Tool — Apollo/Hunter Alt + SMTP Verified", "description": "From $0.0015/lead — 6.7× cheaper than leads-finder $1.5/1k. $0.01/contact for 99%+ run success — SMTP-verified emails + libphonenumber phones + 175+ tech stack. RAG-friendly. AI role + primary-contact scoring. Apollo ($99)/Hunter ($34) alt. x402 + Skyfire-ready. EU/Dutch + GDPR.", "version": "1.9", "x-build-id": "LB4fjkvI7EbitrH6V" }, "servers": [ { "url": "https://api.apify.com/v2" } ], "paths": { "/acts/harvestlab~contact-extractor/run-sync-get-dataset-items": { "post": { "operationId": "run-sync-get-dataset-items-harvestlab-contact-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for its completion, and returns Actor's dataset items in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } }, "/acts/harvestlab~contact-extractor/runs": { "post": { "operationId": "runs-sync-harvestlab-contact-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor and returns information about the initiated run in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK", "content": { "application/json": { "schema": { "$ref": "#/components/schemas/runsResponseSchema" } } } } } } }, "/acts/harvestlab~contact-extractor/run-sync": { "post": { "operationId": "run-sync-harvestlab-contact-extractor", "x-openai-isConsequential": false, "summary": "Executes an Actor, waits for completion, and returns the OUTPUT from Key-value store in response.", "tags": [ "Run Actor" ], "requestBody": { "required": true, "content": { "application/json": { "schema": { "$ref": "#/components/schemas/inputSchema" } } } }, "parameters": [ { "name": "token", "in": "query", "required": true, "schema": { "type": "string" }, "description": "Enter your Apify token here" } ], "responses": { "200": { "description": "OK" } } } } }, "components": { "schemas": { "inputSchema": { "type": "object", "properties": { "urls": { "title": "Website URLs", "type": "array", "description": "List of full website URLs to analyze (e.g. https://example.com). You can provide either URLs or domains.", "items": { "type": "string" } }, "startUrls": { "title": "Start URLs (alias for Website URLs)", "type": "array", "description": "Alternative field name for Website URLs, provided for compatibility with Apify's common 'startUrls' convention (e.g. https://example.com). The canonical field is `urls` above — prefer that. If both are supplied, they are merged.", "items": { "type": "string" } }, "url": { "title": "URL (CLI alias)", "type": "string", "description": "CLI alias for a single website URL. Hidden from Console form." }, "website": { "title": "Website (CLI alias)", "type": "string", "description": "CLI alias for a single website URL. Hidden from Console form." }, "domains": { "title": "Domains", "type": "array", "description": "List of domains without protocol (e.g. example.com). HTTPS will be added automatically. Alternative to providing full URLs — use this OR Website URLs above.", "items": { "type": "string" } }, "maxPagesPerSite": { "title": "Max Pages Per Site", "minimum": 1, "maximum": 20, "type": "integer", "description": "Maximum number of pages to crawl per website. Higher values find more contacts but take longer.", "default": 5 }, "includeSubpages": { "title": "Include Subpages", "type": "boolean", "description": "Automatically crawl /contact, /about, /imprint, and similar pages to find additional contact information.", "default": true }, "detectTechStack": { "title": "Detect Tech Stack", "type": "boolean", "description": "Analyze the website to detect technologies used (CMS, frameworks, analytics, etc.).", "default": true }, "verifyEmails": { "title": "Verify Email Deliverability (MX-record check)", "type": "boolean", "description": "Look up real MX records for every extracted email, detect common-domain typos (e.g. gmial.com -> gmail.com), flag free/disposable inboxes, and tag each email HIGH / MEDIUM / LOW / UNKNOWN deliverability. Adds a new `emails_verified` field to every output item. Cost: $0.01 per email whose domain passes MX resolution (uncertain / disposable / no-MX emails are free).", "default": false }, "deepEmailVerification": { "title": "Deep SMTP Probe — Real Mailbox Verification (v1.7)", "type": "boolean", "description": "Opt-in: run an async SMTP RCPT TO probe against the primary MX host for every MX-cleared email. Classifies each email as deliverable / undeliverable / catchall / greylisted / port_blocked. Reputation-safe: per-MX-host probe history persists across runs (named KV store), capped at 10 probes/host/day with 10s cooldown between probes to the same host. Port 25 is frequently blocked on Apify datacenter egress; when it is, every entry returns `port_blocked` (free). Cost: $0.02 per email that receives a definitive verdict (deliverable or undeliverable) — catch-all, greylisted, and non-verdict results are free. Requires `verifyEmails: true`.", "default": false }, "enableAiAnalysis": { "title": "Enable AI Contact Enrichment", "type": "boolean", "description": "Use an LLM to classify each email by role (sales/support/hr/legal/executive/general/personal), group near-duplicate team addresses, flag non-monitored no-reply inboxes, and pick the single best primary contact for B2B outreach. When `verifyEmails` is also enabled, the AI picks the primary contact only from HIGH/MEDIUM deliverability emails. Requires an API key for your chosen LLM provider. Cost: $0.05 per website analyzed.", "default": false }, "llmProvider": { "title": "LLM Provider", "enum": [ "openrouter", "anthropic", "google", "openai", "ollama" ], "type": "string", "description": "AI backend for email role classification, dedup, and primary-contact selection. 'OpenRouter' (default) is cheapest — Gemini Flash via OpenRouter is ~$0.001 per website. 'Anthropic' = Claude, 'Google AI' = Gemini direct, 'OpenAI' = GPT-4o mini, 'Ollama' = self-hosted (no API cost). Each provider needs its own API key field below.", "default": "openrouter" }, "llmModel": { "title": "LLM Model", "type": "string", "description": "Specific model to use. Leave empty for the provider default (google/gemini-2.0-flash-001 for OpenRouter, claude-sonnet-4-20250514 for Anthropic, gemini-2.0-flash for Google AI, gpt-4o-mini for OpenAI, llama3.1 for Ollama)." }, "openrouterApiKey": { "title": "OpenRouter API Key", "type": "string", "description": "Your OpenRouter API key. Get one at openrouter.ai/keys" }, "anthropicApiKey": { "title": "Anthropic API Key", "type": "string", "description": "Your Anthropic API key. Get one at console.anthropic.com" }, "googleApiKey": { "title": "Google AI API Key", "type": "string", "description": "API key for Google AI (Gemini). Get one at aistudio.google.com/app/apikey" }, "openaiApiKey": { "title": "OpenAI API Key", "type": "string", "description": "API key from platform.openai.com (required if using OpenAI provider)" }, "ollamaBaseUrl": { "title": "Ollama Base URL", "type": "string", "description": "Base URL for Ollama API. Default: http://localhost:11434" }, "proxyConfiguration": { "title": "Proxy Configuration", "type": "object", "description": "Proxy settings. Recommended for reliable results — many websites block requests from datacenter IPs." } } }, "runsResponseSchema": { "type": "object", "properties": { "data": { "type": "object", "properties": { "id": { "type": "string" }, "actId": { "type": "string" }, "userId": { "type": "string" }, "startedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "finishedAt": { "type": "string", "format": "date-time", "example": "2025-01-08T00:00:00.000Z" }, "status": { "type": "string", "example": "READY" }, "meta": { "type": "object", "properties": { "origin": { "type": "string", "example": "API" }, "userAgent": { "type": "string" } } }, "stats": { "type": "object", "properties": { "inputBodyLen": { "type": "integer", "example": 2000 }, "rebootCount": { "type": "integer", "example": 0 }, "restartCount": { "type": "integer", "example": 0 }, "resurrectCount": { "type": "integer", "example": 0 }, "computeUnits": { "type": "integer", "example": 0 } } }, "options": { "type": "object", "properties": { "build": { "type": "string", "example": "latest" }, "timeoutSecs": { "type": "integer", "example": 300 }, "memoryMbytes": { "type": "integer", "example": 1024 }, "diskMbytes": { "type": "integer", "example": 2048 } } }, "buildId": { "type": "string" }, "defaultKeyValueStoreId": { "type": "string" }, "defaultDatasetId": { "type": "string" }, "defaultRequestQueueId": { "type": "string" }, "buildNumber": { "type": "string", "example": "1.0.0" }, "containerUrl": { "type": "string" }, "usage": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "integer", "example": 1 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } }, "usageTotalUsd": { "type": "number", "example": 0.00005 }, "usageUsd": { "type": "object", "properties": { "ACTOR_COMPUTE_UNITS": { "type": "integer", "example": 0 }, "DATASET_READS": { "type": "integer", "example": 0 }, "DATASET_WRITES": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_READS": { "type": "integer", "example": 0 }, "KEY_VALUE_STORE_WRITES": { "type": "number", "example": 0.00005 }, "KEY_VALUE_STORE_LISTS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_READS": { "type": "integer", "example": 0 }, "REQUEST_QUEUE_WRITES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_INTERNAL_GBYTES": { "type": "integer", "example": 0 }, "DATA_TRANSFER_EXTERNAL_GBYTES": { "type": "integer", "example": 0 }, "PROXY_RESIDENTIAL_TRANSFER_GBYTES": { "type": "integer", "example": 0 }, "PROXY_SERPS": { "type": "integer", "example": 0 } } } } } } } } } } ```