Linkedin Harvester MCP Server

LinkedIn Profile Finder - Email to LinkedIn URL

Turn an email address into the person's public LinkedIn profile, with a confidence score on every match. Paste one email or a whole list and get back a clean table of LinkedIn URLs for your CRM, outreach or recruiting pipeline. No LinkedIn login, no cookies, nothing tied to your account.

Available as an Apify Actor. $0.005 per email checked plus $0.02 per confident match. Works with no API keys; free Brave Search and Groq keys widen coverage.

What you get

• Email in, best-match LinkedIn URL out, with a 0 to 100% confidence score on every row so you know which matches to trust.
• Bulk or one-off. A single address or thousands, one result row each.
• Pay for results. Half a cent per email checked; the match fee applies only when a URL clears your confidence gate.
• Public information only. GitHub commit authors, Gravatar, PGP keyservers and handle probes on 8 platforms. No data broker, no private database.
• Account-safe. It never logs into LinkedIn or touches your session, so there is nothing to get flagged.

Who it's for

• Sales and growth: enrich inbound leads and sign-ups with the right LinkedIn profile before you reach out.
• Recruiting: turn a list of candidate emails into LinkedIn profiles.
• RevOps / CRM: clean, verify and backfill contact records at scale.
• Founders and operators: see who actually emailed you.

How to use it

1. Paste a single email, or a list of emails. The form comes prefilled with two demo addresses; replace them with yours.
2. Leave Check public profiles on. It recovers the name and a cross-platform handle, which is what resolves a URL without a search key.
3. Pick a matching mode. Balanced is the default and works for most lists.
4. Optional: add your free Brave Search and Groq keys for web-search corroboration and AI name-splitting.
5. Hit Start. Results appear in the dataset, one row per email.

Matching modes

What does the output look like?

This is the real dataset from a run on 2026-09-12 (build 0.1.5, no API keys, default settings). Six emails, 25 seconds, 1 GB of memory.

Confident means the row cleared the 80% gate. Those are the only rows that carry the match fee. Rows between 50% and 79% are possible matches, returned free with their score, for you to verify before use.

One row as JSON (the signals block is what the scorer saw):

{
  "email": "anshumanatrey@gmail.com",
  "found": true,
  "linkedin_url": "https://www.linkedin.com/in/anshumanatrey",
  "confidence": 0.97,
  "passes_gate": true,
  "gate_threshold": 0.8,
  "derived_name": "Anshuman Atrey",
  "name_source": "github_commit",
  "source": "handle_pivot",
  "evidence": "handle 'anshumanatrey' confirmed on 4 platform(s) -> likely LinkedIn slug",
  "signals": {
    "names": [{"value": "Anshuman Atrey", "source": "github_commit"}],
    "handles": ["anshumanatrey"],
    "confirmed_socials": ["GitHub: https://github.com/anshumanatrey", "YouTube: https://www.youtube.com/@anshumanatrey", "Telegram: https://t.me/anshumanatrey"]
  }
}

What does it cost?

Pay per event. Platform compute is included in the prices.

Typical jobs:

The 6-email run used 0.0069 compute units at the 1 GB default, about 4 seconds per email.

How good are the matches?

Measured on the run above: 5 of 6 addresses resolved to a URL, 1 above the gate, and the role mailbox was correctly skipped.

• Work emails (first.last@company.com) give the name from the address itself. The handle probes then look for that name as a handle across GitHub, GitLab, dev.to, npm, Gravatar, Linktree, About.me, YouTube and Telegram. Three or more hits put the match at 78%.
• Personal addresses resolve when the person has a public developer or social footprint. A GitHub commit author name or a Gravatar display name is authoritative and lifts confidence to the high 90s.
• Over the gate. A confident match needs a name from an authoritative source plus a handle that is consistent across platforms. A free Brave Search key adds a web-search check on each candidate, the designed way to lift 78% matches over the gate.
• Not found is a clean not found, never a confident guess. Role mailboxes (info@, support@, sales@) are skipped rather than probed.

Good to know

• Speed: about 4 seconds per email at the 1 GB default; a 100-email list finishes in about 7 minutes.
• Works on public information only. No private database, no broker-sourced contact data.
• LinkedIn login-walls its pages, so a URL is a strong lead confirmed by cross-platform evidence, not a fetched profile. Turn on Scrape the matched profile with a Bright Data key if you need the full profile.
• You stay in control. Set the confidence bar, choose how much AI to use, bring your own keys.

Sibling actors

Run it yourself (open source)

The actor is open source under the MIT license. To run it from the command line with your own keys:

git clone https://github.com/AnshumanAtrey/linkedin-harvester.git
cd linkedin-harvester
pip3 install -r requirements.txt

GROQ_API_KEY=your_groq_key BRAVE_API_KEY=your_brave_key \
  python3 -m harvester.find "satya.nadella@microsoft.com"

Free keys: Groq at console.groq.com/keys, Brave at brave.com/search/api

Use responsibly

Built for legitimate sales, recruiting, research and verification. Follow the privacy laws that apply to you (GDPR, CCPA and local equivalents) and honor opt-out and do-not-contact requests.

License

MIT, see LICENSE.

Last updated

2026-09-12 (version 0.1.5)