linkedin-contact-reacher

name: linkedin-contact-reacher description: Systematically discover LinkedIn contacts for job referrals or networking using BFS/DFS traversal. Extracts LinkedIn URLs, generates company email candidates, optionally sends connection requests, and saves output locally as JSON + CSV.

LinkedIn Contact Reacher Skill

Discover LinkedIn contacts systematically (BFS or DFS), enrich them with email candidates, optionally send personalized connection requests, and save everything locally.

Specialized Agents

Script Files

Prerequisites

• Logged into LinkedIn
• Playwright MCP tools available
• Output directory writable (default: ./output/)

Orchestrated Run Order

1. QA Agent          → preFlightCheck(page)
2. Discovery Agent   → discoverContacts()      ← BFS or DFS
3. Extract Agent     → extractContactInfo()    ← enrich + email candidates
4. [Optional] Reach  → reachContacts()         ← send connection requests
5. Save              → saveOutput()            ← JSON + CSV locally
6. QA Agent          → generateReport()

Quick Start

Scenario 1: Find referral contacts at a target company

// ── Step 1: Discover contacts via BFS ─────────────────────────────────
// Paste discoverContacts.js, then:

const contacts = await discoverContacts(page, {
  seeds: [
    { type: 'search', company: 'Google', role: 'Engineering Manager' },
    { type: 'search', company: 'Google', role: 'Software Engineer' }
  ],
  strategy: 'bfs',          // breadth-first: broad coverage of the company
  maxContacts: 30,
  maxDepth: 1,              // depth=1: only direct search results, no expansion
  targetCompanies: ['Google'],
  targetRoles: ['engineer', 'manager', 'tech lead'],
  connectionDegree: ['1st', '2nd', '3rd'],
});

// ── Step 2: Enrich with email candidates ──────────────────────────────
// Paste extractContactInfo.js, then:

const enriched = await extractContactInfo(page, contacts, {
  companyDomains: { 'Google': 'google.com' },
  guessPersonalEmail: false,   // set true to also get gmail/outlook guesses
});

// ── Step 3: Save output ───────────────────────────────────────────────
// Paste saveOutput.js, then:

await saveOutput(enriched, {
  format: 'both',
  outputDir: './output',
  label: 'google-referrals'
});
// Output: ./output/google-referrals_2026-04-27T....{json,csv}

// ── Step 4: (Optional) Send connection requests ───────────────────────
// Paste reachContacts.js, then:

const { results, stats } = await reachContacts(page, enriched, {
  purpose: 'referral',
  userProfile: {
    name: 'Your Name',
    role: 'Software Engineer',
    targetCompany: 'Google'
  },
  maxPerSession: 10,      // stay under LinkedIn's limit
});

// Save final results with reach status
await saveOutput(results, { label: 'google-referrals-reached', format: 'both' });

Scenario 2: Networking DFS — deep dive from one person's profile

const contacts = await discoverContacts(page, {
  seeds: [{ type: 'profile', url: 'https://www.linkedin.com/in/someengineer/' }],
  strategy: 'dfs',          // depth-first: explore one thread deeply
  maxContacts: 20,
  maxDepth: 2,
  targetCompanies: ['Stripe', 'Shopify', 'Airbnb'],
  connectionDegree: ['2nd', '3rd'],
});

const enriched = await extractContactInfo(page, contacts, {
  companyDomains: {
    'Stripe':   'stripe.com',
    'Shopify':  'shopify.com',
    'Airbnb':   'airbnb.com',
  },
  guessPersonalEmail: false,
});

await saveOutput(enriched, { label: 'networking-dfs', format: 'both' });

Scenario 3: Multiple companies, networking only (no outreach)

const allContacts = [];

for (const target of [
  { company: 'Stripe', role: 'backend engineer' },
  { company: 'Shopify', role: 'software engineer' },
]) {
  const batch = await discoverContacts(page, {
    seeds: [{ type: 'search', ...target }],
    strategy: 'bfs',
    maxContacts: 15,
    maxDepth: 1,
    targetCompanies: [target.company],
  });
  allContacts.push(...batch);
  await page.waitForTimeout(5000); // pause between companies
}

const enriched = await extractContactInfo(page, allContacts, {
  companyDomains: { 'Stripe': 'stripe.com', 'Shopify': 'shopify.com' }
});

await saveOutput(enriched, { label: 'multi-company', format: 'both' });

Configuration Reference

discoverContacts(page, options)

extractContactInfo(page, contacts, options)

reachContacts(page, contacts, options)

saveOutput(contacts, options)

Output Schema

Each contact in the output has:

{
  "name": "Jane Smith",
  "firstName": "Jane",
  "lastName": "Smith",
  "title": "Engineering Manager",
  "company": "Google",
  "location": "San Francisco, CA",
  "linkedinUrl": "https://www.linkedin.com/in/janesmith/",
  "connectionDegree": "2nd",
  "depth": 0,
  "companyDomain": "google.com",
  "emailCandidates": [
    "jane.smith@google.com",
    "jsmith@google.com",
    "janesmith@google.com",
    "janes@google.com"
  ],
  "personalEmailCandidates": [],
  "publicEmail": null,
  "websites": [],
  "messageSent": false,
  "reachStatus": "",
  "messageText": "",
  "reachedAt": null,
  "via": "seed:search"
}

BFS vs DFS — When to Use Which

Email Candidate Patterns

Generated email patterns (in priority order):

Verify candidates with tools like Hunter.io or email verification APIs before sending.

Safety & Rate Limits

LinkedIn may restrict accounts that send too many requests. Start conservatively.

Legal & Ethical Usage

• Only contact people you have a genuine professional reason to reach
• Personalize messages — do not spam
• Respect LinkedIn's Terms of Service
• Do not use harvested emails for bulk cold outreach without consent
• Email candidates are guesses only — verify before use