AI Email Outreach System
that Closes Deals While You Sleep

# Instantly API Skill

## Purpose
Manage cold email campaigns via Instantly.ai API.
Create campaigns, add leads, check stats, pause/resume.

## API Base URL
https://api.instantly.ai/api/v1

## Authentication
All requests require: ?api_key=$INSTANTLY_API_KEY

## Available Operations
- list_campaigns: GET /campaign/list
- create_campaign: POST /campaign/create
- add_leads: POST /lead/add
- get_campaign_analytics: GET /analytics/campaign/summary
- check_replies: GET /unibox/emails (filter: is_reply=true)
- pause_campaign: POST /campaign/pause
- resume_campaign: POST /campaign/resume

## Script Reference
See ./instantly.js for full implementation

// Instantly API Helper — OpenClaw Skill
const BASE_URL = 'https://api.instantly.ai/api/v1';
const API_KEY = process.env.INSTANTLY_API_KEY;

async function addLeadToCampaign({ campaignId, email, firstName, lastName, companyName, personalization }) {
  const res = await fetch(`${BASE_URL}/lead/add?api_key=${API_KEY}`, {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      campaign_id: campaignId,
      leads: [{
        email,
        first_name: firstName,
        last_name: lastName,
        company_name: companyName,
        personalization  // Custom variable for email template
      }]
    })
  });
  return res.json();
}

async function getCampaignStats(campaignId) {
  const res = await fetch(
    `${BASE_URL}/analytics/campaign/summary?api_key=${API_KEY}&campaign_id=${campaignId}`
  );
  return res.json();
}

async function checkNewReplies(since) {
  const res = await fetch(
    `${BASE_URL}/unibox/emails?api_key=${API_KEY}&is_reply=true&from_date=${since}`
  );
  return res.json();
}

module.exports = { addLeadToCampaign, getCampaignStats, checkNewReplies };

# Lead Research Skill

## Purpose
Research prospect companies and contacts to gather
personalization data for cold outreach.

## Research Sequence
For each lead, execute in order:

1. Hunter.io domain search → find verified emails
   GET https://api.hunter.io/v2/domain-search?domain={domain}&api_key=$HUNTER_API_KEY

2. Apollo.io people search → enrich contact data
   POST https://api.apollo.io/v1/people/search

3. Browser tool → scrape company LinkedIn page
   - Extract: recent posts, team size changes, job listings
   - Look for: hiring signals, product launches, funding news

4. Browser tool → Google news search
   Query: "{company_name} news site:techcrunch.com OR site:businesswire.com"
   - Extract last 3 months of press mentions

## Output Format
Return structured JSON:
{
  "contact": { "name", "email", "title", "linkedin_url" },
  "company": { "name", "domain", "size", "industry", "tech_stack" },
  "triggers": ["raised Series B", "hiring 5 engineers", "new CPO joined"],
  "pain_points": ["manual processes", "scaling challenges"],
  "personalization_hook": "Saw you just raised your Series B — congrats!"
}

## ICP Matching
Before researching, check ~/.openclaw/memory/icp.md
Only research leads that match ICP criteria.
Score each lead: High / Medium / Low fit.
Skip Low fit leads automatically.

# Email Personalization Skill

## Purpose
Write hyper-personalized cold emails that get replies.
Each email must feel hand-written, not templated.

## Email Structure
- Subject: 3–6 words, curiosity-driven, no spam words
- Opening line: Reference THE SPECIFIC TRIGGER (required)
- Problem statement: 1 sentence connecting trigger → pain
- Value proposition: 1 sentence, concrete outcome
- Social proof: 1 micro-proof (optional, keep it brief)
- CTA: One soft ask, frictionless

## Rules
- Max 120 words for initial email
- NO: "I hope this finds you well", "I wanted to reach out"
- NO: Generic compliments ("love your company")
- YES: Specific, verifiable triggers
- YES: First name only, conversational tone
- YES: End with question or ultra-low-friction ask

## Personalization Variables for Instantly
Use {{personalization}} variable → inject unique opening line per lead
Use {{first_name}}, {{company_name}} for Instantly templates

## Follow-up Sequence Logic
Day 3: Angle 2 (different value prop)
Day 7: Case study / social proof angle  
Day 14: Break-up email ("closing your file")
Stop sequence immediately on any reply detected.

// Reply Monitor — runs every 30 min via OpenClaw cron
const { checkNewReplies } = require('../instantly/instantly');

async function monitorReplies() {
  const since = new Date(Date.now() - 30 * 60 * 1000).toISOString();
  const replies = await checkNewReplies(since);

  for (const reply of replies.data || []) {
    // Classify reply sentiment with AI
    const sentiment = await classifyReply(reply.body);

    // Send Telegram alert with context
    await telegramAlert({
      message: `🔥 New Reply!\n\n` +
        `From: ${reply.from_email}\n` +
        `Company: ${reply.from_company}\n` +
        `Sentiment: ${sentiment}\n\n` +
        `"${reply.body.slice(0, 200)}..."\n\n` +
        `👉 Suggested response: [Aria drafts reply]`
    });
  }
}

// Tell Aria via Telegram to set this up:
// "Run reply monitor every 30 minutes and alert me on Telegram"

🦞 Aria Morning Briefing — Monday, Feb 24

📤 Yesterday's Activity
  • Researched: 15 leads
  • ICP High fit: 6 | Medium: 7 | Skipped: 2
  • Added to campaign: 13 leads
  • Emails sent (Instantly): 13 initial + 8 follow-ups

🔔 Replies (overnight)
  • 2 new replies detected
  • HubTech Inc. (Sarah M., VP HR): POSITIVE — wants to chat
  • Growly SaaS (Mark T.): OUT_OF_OFFICE — auto-paused 5 days

📈 Campaign Stats (all-time)
  • Total sent: 284
  • Open rate: 51.2% ✅
  • Reply rate: 11.3% ✅  
  • Meetings booked: 8 🎉

📋 Today's Queue
  • 10 new leads in ~/leads/queue.csv
  • Processing starts in 2 hours
  • Tap ✅ to approve auto-send, or ❌ to review first