name: linkedin-adapter
description: |
LinkedIn automation adapter for the 100X Outreach System.
Use this skill when performing LinkedIn actions like connecting, messaging, viewing profiles, liking posts, or commenting.
This skill controls the browser extension to execute LinkedIn actions with templates.
allowed-tools:
- Bash
- Read
- Write
- Glob
- Grep
- TodoWrite
- AskUserQuestion
- WebFetch

LinkedIn Adapter Skill

Automates LinkedIn actions using the browser extension with intelligent template rendering.

Browser Extension: Local Control

Fast, local browser automation! The extension provides direct control over your browser via WebSocket.

Benefits:
- ✅ Faster: Local execution, no cloud latency
- ✅ Free: No usage costs
- ✅ Visible: See actions in real-time
- ✅ Persistent: Uses your real LinkedIn account
- ✅ Activity Tracking: Built-in LinkedIn Lookback integration
- ✅ Rate Limiting: Smart rate limits prevent LinkedIn detection

Architecture

Claude Code (You)
    ↓ HTTP API
Canvas Server (localhost:3000)
    ↓ WebSocket
Browser Extension
    ↓ Chrome APIs
LinkedIn.com

IMPORTANT: You Control the Extension

You are the brain that orchestrates the extension. This skill tells you exactly how to:
1. Load and render templates
2. Send commands to the extension via HTTP API
3. Wait for action results
4. Monitor and report progress

When to Use This Skill

Use this skill when the user wants to:
- Send LinkedIn connection requests
- Message LinkedIn connections
- View LinkedIn profiles (warm-up)
- Like/comment on LinkedIn posts
- Send InMails

Available LinkedIn Actions

Available Templates

Connection Requests (templates/linkedin/connection-requests/)

• cold_outreach.md - General cold outreach
• startup_founder.md - Reach out to founders
• thought_leader.md - Connect with thought leaders
• speaker_invitation.md - Invite speakers
• content_collaboration.md - Content collab requests
• alumni_connect.md - Alumni connections
• potential_client.md - Sales outreach
• mutual_connection.md - Mutual connection intro
• same_industry.md - Same industry connection
• recruiter.md - Recruiter outreach
• investor_outreach.md - Investor connections
• event_attendee.md - Event attendee connections

Messages (templates/linkedin/messages/)

• intro_after_connect.md - First message after connecting
• follow_up_no_response.md - Follow up on no response
• meeting_request.md - Request a meeting
• collaboration_proposal.md - Propose collaboration
• value_offer.md - Offer value
• thank_you.md - Thank you message
• partnership_proposal.md - Partnership proposals
• sales_intro.md - Sales introduction
• referral_request.md - Ask for referrals
• expertise_request.md - Ask for expert advice
• podcast_invitation.md - Podcast guest invitation

InMails (templates/linkedin/inmails/)

• cold_inmail.md - Cold InMail
• executive_outreach.md - Executive outreach
• job_opportunity.md - Job opportunities

Comments (templates/linkedin/comments/)

• thoughtful_engagement.md - Thoughtful comment
• question_comment.md - Ask a question
• congratulations.md - Congratulate them

CRITICAL: Step-by-Step Execution Flow

Prerequisites

1. Canvas server must be running:
   bash cd canvas && npm start # Server at http://localhost:3000 # WebSocket at ws://localhost:3000/ws

2. Browser extension must be installed:

3. Load .claude/skills/browser-extension in Chrome

4. Extension badge should show "✓" (connected)

5. User must be logged into LinkedIn in their browser

Step 1: Check Extension Connection

Verify the extension is connected:

curl http://localhost:3000/api/extension/status

Expected response:

{
  "connected": true,
  "extensionId": "abc123...",
  "capabilities": ["linkedin", "instagram", "twitter", "google"]
}

If not connected, tell user to check extension popup.

Step 2: Load and Render Template

Read the template file and render with Jinja2:

# List available templates
python .claude/scripts/template_loader.py list --platform linkedin --category connection-requests

# Render a template with variables
python .claude/scripts/template_loader.py render --path linkedin/connection-requests/cold_outreach --var first_name "John" --var company "Acme Inc" --var industry "AI/ML" --var my_name "Your Name"

This returns the rendered message text.

Step 3: Send Action to Extension

Send LinkedIn action command via HTTP API:

# Connection request
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "send_connection",
    "profileUrl": "https://linkedin.com/in/johnsmith",
    "note": "Hi John, I came across your profile..."
  }'

# Send message
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "send_message",
    "profileUrl": "https://linkedin.com/in/johnsmith",
    "message": "Thanks for connecting! I wanted to reach out..."
  }'

# View profile (warm-up)
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "view_profile",
    "profileUrl": "https://linkedin.com/in/johnsmith"
  }'

# Like post
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "like_post",
    "postUrl": "https://linkedin.com/feed/update/urn:li:activity:123456"
  }'

# Comment on post
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "comment",
    "postUrl": "https://linkedin.com/feed/update/urn:li:activity:123456",
    "comment": "Great insights! I especially liked..."
  }'

# Send InMail (requires Premium)
curl -X POST http://localhost:3000/api/linkedin/action \
  -H "Content-Type: application/json" \
  -d '{
    "type": "send_inmail",
    "profileUrl": "https://linkedin.com/in/johnsmith",
    "subject": "Partnership Opportunity",
    "body": "Hi John, I have an exciting partnership opportunity..."
  }'

Step 4: Wait for Result

The API call returns immediately with the result:

Success Response:

{
  "success": true,
  "action": "send_connection",
  "timestamp": "2026-01-22T17:30:00Z",
  "status": "sent",
  "note": "Hi John, I came across your profile...",
  "rateLimit": {
    "allowed": true,
    "remaining": 14,
    "limit": 15,
    "resetDate": "2026-01-22"
  }
}

Error Response:

{
  "success": false,
  "action": "send_connection",
  "error": "Daily connection limit reached (15/day). Resets tomorrow.",
  "timestamp": "2026-01-22T17:30:00Z"
}

Step 5: Report to User

Tell the user what happened:
- ✅ "Successfully sent connection request to John Smith (14 connections remaining today)"
- ❌ "Failed to send connection: Daily limit reached. Resets tomorrow."
- ⏸️ "Profile already connected. Skipping."

Step 6: Record Activity (Automatic)

The extension automatically records all LinkedIn activity:
- Tracks profile views, connections, messages
- Stores in browser IndexedDB
- Forwards to LinkedIn Lookback database
- Triggers automated follow-up workflows

No manual recording needed!

Example: Send Connection Request

User: "Connect with John Smith on LinkedIn with a cold outreach message"

You should:

1. Check extension:
   bash curl http://localhost:3000/api/extension/status
   → Verify extension is connected

2. Render template:
   bash python .claude/scripts/template_loader.py render --path linkedin/connection-requests/cold_outreach --var first_name "John" --var company "UNKNOWN" --var my_name "Your Name"
   → Get rendered message: "Hi John, I came across your profile..."

3. Send to extension:
   bash curl -X POST http://localhost:3000/api/linkedin/action \ -H "Content-Type: application/json" \ -d '{ "type": "send_connection", "profileUrl": "https://linkedin.com/in/johnsmith", "note": "Hi John, I came across your profile..." }'

4. Get result (immediate):
   json { "success": true, "status": "sent", "rateLimit": { "remaining": 14, "limit": 15 } }

5. Report to user:
   ✅ Connected with John Smith successfully! (14 connections remaining today)

That's it! The extension handles all the browser automation automatically.

Example: Bulk Outreach (Single Approval)

User: "Connect with these 5 LinkedIn profiles"

You should:

1. Show preview:
   ```
   ═══════════════════════════════════════════
   LINKEDIN OUTREACH - 5 CONNECTIONS
   ═══════════════════════════════════════════

Template: Cold Outreach

Targets:
1. John Smith (CEO at Acme)
2. Jane Doe (CTO at TechCo)
3. Bob Wilson (Founder at StartupX)
4. Alice Chen (VP at BigCorp)
5. Mike Brown (Director at MidCo)

Rate Limit: 15 connections remaining today
Estimated time: ~15 minutes (3 min/connection)
═══════════════════════════════════════════

Proceed with all 5 connection requests?
```

1. After single approval, execute ALL actions autonomously:
   ```bash
   for person in targets:
   # 1. Render template
   message = render_template(person)

   # 2. Send to extension
   result = curl POST /api/linkedin/action {
   "type": "send_connection",
   "profileUrl": person.url,
   "note": message
   }

   # 3. Report progress
   print(f"✅ {person.name}: {result.status}")

   # 4. Smart delay (2-5 minutes between connections)
   sleep(random(120, 300))
   ```

2. Final summary:
   ```
   ✅ Bulk outreach complete!

3. 5/5 connections sent successfully
4. 10 connections remaining today
5. Activity recorded in LinkedIn Lookback
   ```

Rate Limits (Built into Extension)

The extension automatically enforces these daily limits:

All limits reset at midnight (local time).

Check Rate Limits

# Check extension status and rate limits
curl http://localhost:3000/api/linkedin/limits

Response:

{
  "connections": { "used": 5, "remaining": 10, "limit": 15 },
  "messages": { "used": 12, "remaining": 28, "limit": 40 },
  "profileViews": { "used": 45, "remaining": 55, "limit": 100 },
  "likes": { "used": 20, "remaining": 30, "limit": 50 },
  "comments": { "used": 5, "remaining": 25, "limit": 30 },
  "inmails": { "used": 0, "remaining": 5, "limit": 5 },
  "resetDate": "2026-01-23"
}