LinkedIn MCP by GTM API¶
The first production-grade, ban-resistant LinkedIn MCP server. Managed by GTM API, this server handles the entire anti-bot stack — rotating residential IPs, browser fingerprinting, session management, and CAPTCHA solving — so your AI agent can focus on LinkedIn actions, not infrastructure.
What It Does¶
- Search: Find people and companies on LinkedIn with filters (industry, role, location, company size)
- Connect: Send connection requests with personalized notes
- Message: Send InMail and direct messages through LinkedIn Messaging
- Enrich: Pull full profile data, company pages, and relationship graphs
- Track: Monitor profile views, post engagement, and connection acceptance rates
Why It Matters for Operators¶
Before this MCP, AI-driven LinkedIn outreach meant: 1. Human copies a lead list from Apollo/ZoomInfo 2. Human logs into LinkedIn and manually connects/messages 3. Human tracks responses in a CRM 4. Human follows up
Now: AI agent searches LinkedIn → enriches with Apollo.io data → sends connection request → follows up with message → logs to CRM — all in one workflow.
Setup¶
Prerequisites¶
- GTM API account (sign up at gtmapi.com)
- API key from GTM API dashboard
Claude Desktop¶
{
"mcpServers": {
"linkedin-gtm": {
"type": "streamable-http",
"url": "https://api.gtmapi.com/mcp/linkedin",
"headers": {
"Authorization": "Bearer YOUR_GTM_API_KEY"
}
}
}
}
Cursor / VS Code¶
{
"mcpServers": {
"linkedin-gtm": {
"type": "streamable-http",
"url": "https://api.gtmapi.com/mcp/linkedin",
"headers": {
"Authorization": "Bearer YOUR_GTM_API_KEY"
}
}
}
}
Hermes Agent¶
# ~/.hermes/config.yaml
mcp_servers:
linkedin-gtm:
type: streamable-http
url: https://api.gtmapi.com/mcp/linkedin
headers:
Authorization: "Bearer ${GTM_API_KEY}"
Tools¶
| Tool | Description |
|---|---|
search_people |
Search LinkedIn profiles by keyword, title, company, location |
get_profile |
Get full profile details by LinkedIn URL or profile ID |
send_connection_request |
Send a connection request with optional note |
send_message |
Send a direct message to a 1st-degree connection |
get_messages |
Retrieve recent conversation threads |
search_companies |
Search company pages by name, industry, size |
get_company |
Get full company page details |
enrich_email |
Find LinkedIn profile from email address |
get_network_graph |
Get mutual connections and relationship paths |
check_connection_status |
Check connection degree to any profile |
Pricing¶
| Tier | Price | Includes |
|---|---|---|
| Starter | Free | 100 searches/month, 50 connection requests |
| Growth | $49/mo | 1,000 searches, 500 connection requests, messaging |
| Scale | $199/mo | 10,000 searches, 5,000 connection requests, enrichment |
| Enterprise | Custom | Unlimited, dedicated IP pool, priority support |
Anti-Ban Architecture¶
GTM API's infrastructure includes: - Rotating residential IPs: 50K+ IPs across 195 countries — no datacenter IP blocks - Browser fingerprinting: Real Chrome fingerprints with consistent WebGL, canvas, and font profiles - Session management: Persistent sessions with human-like activity patterns (random delays, scroll behavior) - CAPTCHA solving: Automatic FunCaptcha and reCAPTCHA resolution - Rate limiting: Intelligent backoff based on LinkedIn's current enforcement posture
This is why they achieve <1% ban rate across 20,000+ accounts — operators don't need to understand any of this. The AI agent just calls tools.
Use Cases¶
B2B Lead Generation¶
Agent: "Find VP of Engineering at Series B SaaS companies in the Bay Area with 50-200 employees who posted about AI in the last month"
→ search_people → enrich with Apollo.io → send_connection_request → follow up with message
Competitive Research¶
Agent: "Who are the top salespeople at [competitor] and what are they posting about?"
→ search_people at competitor → get_profile for each → analyze recent posts
Investor Outreach¶
Agent: "Find partners at AI-focused VC firms who invested in devtools in the last 12 months"
→ search_people by title "Partner" at VC firms → get_profile → cross-reference portfolio
Talent Sourcing¶
Agent: "Find ML engineers at FAANG companies who contribute to PyTorch on GitHub"
→ search_people → get_profile → cross-reference with GitHub activity
Limitations¶
- Not for scraping at scale: GTM API enforces fair-use limits. This is for relationship-building, not data mining.
- LinkedIn's terms of service: Automated LinkedIn activity carries risk. GTM API's managed infrastructure mitigates this, but operators should understand compliance implications.
- 1st-degree messaging only:
send_messagerequires an existing connection. For cold outreach, usesend_connection_requestwith a note first.
See Also¶
- [[apollo-io-mcp]] — B2B contact enrichment (complementary)
- [[x-use-mcp]] — X/Twitter automation (similar approach)
- [[mercury-mcp]] — Banking for operators (fintech stack)