LeadShark
Dashboard

LeadShark Automations MCP

Pro+

Control your LeadShark automations with natural language. No UI required.

Execution-focused. Fail-safe. Intent → Automation.

Overview

The LeadShark MCP (Model Context Protocol) server lets AI assistants like Cursor, Claude, and ChatGPT manage your entire automation workflow through conversation.

6 Tools

End-to-end automation control

Fail-Safe

Draft by default, confirm destructive

No UI

Intent → Automation directly

What You Can Do

list_recent_posts— See your latest LinkedIn posts
create_automation— Create from natural language
list_automations— View all with performance stats
edit_automation— Update status, keywords, copy
schedule_post_with_automation— Schedule + auto-attach automation
set_daily_dm_limit— Safety lever (0 = emergency stop)

What You Can Say

Talk to your AI assistant naturally. It figures out which tool to use.

“Show me my recent posts”

→ list_recent_posts

“Automate commenters on my Claude guide post”

→ create_automation

“DM anyone who says 'interested' and auto-connect with them”

→ create_automation

“List all my running automations”

→ list_automations

“Pause the AI tools automation”

→ edit_automation

“Enable auto-like on my Claude guide automation”

→ edit_automation

“Add a follow-up DM that sends after 2 hours if no reply”

→ edit_automation

“Change the DM template to mention the new pricing”

→ edit_automation

“Turn on partially engage so I don't double-reply”

→ edit_automation

“Enable link tracking on my active automations”

→ edit_automation

“Schedule a post for tomorrow 9am EST”

→ schedule_post_with_automation

“Schedule my Claude guide giveaway post for Monday morning with full automation”

→ schedule_post_with_automation

“Plan a lead magnet post: publish Friday 10am, DM anyone who comments 'interested'”

→ schedule_post_with_automation

“Set my daily DM limit to zero”

→ set_daily_dm_limit(emergency stop)

Tools (6)

Complete automation control. Each tool does one thing well.

Example Workflow

A typical conversation from discovery to optimized automation:

You:

“Show me my recent posts”

AI:

Here are your last 5 posts. The Claude guide post has 142 comments and no automation yet.

→ list_recent_posts

You:

“Automate that one - DM anyone who says interested, and auto-connect with them”

AI:

Created 'Claude Guide Giveaway' as Draft. Keywords: interested. Auto-connect enabled. Ready to review.

→ create_automation

You:

“Looks good, activate it and enable auto-like too”

AI:

Automation is now Running with auto-like enabled. I'll like comments and DM people who say 'interested'.

→ edit_automation

You:

“Add 'send' and 'want' as keywords, and turn on partially engage”

AI:

Updated. Keywords: interested, send, want. Partially engage enabled — won't reply to comments you've already responded to.

→ edit_automation

You:

“Enable link tracking so I can see who clicks”

AI:

Link tracking enabled. Your DM links will use leads.sh for click tracking.

→ edit_automation

You:

“Add a follow-up DM after 2 hours if they don't reply”

AI:

Follow-up enabled. Will send after 2 hours only if no response received.

→ edit_automation

You:

“How's it doing?”

AI:

284 comments, 156 replied, 142 DMs sent, 89 follow-ups sent, 67 new connections. Top keyword: 'interested' (67%).

→ list_automations

You:

“Schedule a new post for Monday 9am EST with the same automation setup”

AI:

Scheduled for Monday 9am EST. Post + automation ready. Automation will activate automatically when post publishes.

→ schedule_post_with_automation

You:

“Pause everything, I'm going on vacation”

AI:

Daily DM limit set to 0. All sending paused across 3 automations. Automations will still capture comments. Set limit back when you return.

→ set_daily_dm_limit

Setup

Requirements

Install

npm install @leadshark/mcp-server

Cursor Configuration

Create .cursor/mcp.json:

{
  "mcpServers": {
    "leadshark": {
      "command": "npx",
      "args": ["-y", "@leadshark/mcp-server"],
      "env": {
        "LEADSHARK_API_KEY": "${LEADSHARK_API_KEY}"
      }
    }
  }
}

Restart Cursor after adding.

Claude Desktop

claude mcp add leadshark \
  --transport stdio \
  --env LEADSHARK_API_KEY=your_key \
  -- npx -y @leadshark/mcp-server
🦞

OpenClaw Integration

Coming Soon

Manage automations via WhatsApp, Telegram, or Discord using OpenClaw.

Safety Model

ActionDefault Behavior
Create automationDraft status, requires activation
Edit automationPartial updates only, no destructive changes
Delete automationNot supported — pause instead
Set DM limit to 0Requires confirm: true
Schedule postAutomation activates only when post publishes

Permission Scoped

All actions use your existing LeadShark permissions

Thin Layer

No background jobs, no side effects, just control

Auditable

Every action maps to a dashboard equivalent

Settings Reference

All available settings you can configure via the MCP:

Engagement Settings

auto_connect— Auto-accept incoming connections & send connection requests to commenters
auto_like— Automatically like all comments on your post
partially_engage— Only engage with comments that don't already have a manual reply from you
link_tracking— Track link clicks using leads.sh short links

Follow-up DM Settings

follow_up_enabled— Enable automated follow-up DMs
follow_up_template— The follow-up message (supports {{firstName}}, {{lastName}}, etc.)
follow_up_delay_minutes— Delay before sending (1 min to 7 days)
follow_up_only_if_no_response— Only send follow-up if lead hasn't replied

Template Variables

Comment Replies (with @mention support):

{{fullNameMention}}★
{{firstNameMention}}
{{firstName}}
{{fullName}}

★ Recommended — @mentions notify the commenter

DMs & Follow-ups (no @mention):

{{firstName}}{{lastName}}{{fullName}}{{linkedinUsername}}

@mentions don't work in DMs — use plain variables

Who This Is For

Power Users

Multiple automations

Agencies

Client automation at scale

Developers

AI-native workflows

Founders

Intent → Automation, fast

Questions?

Reach out for MCP support or feature requests.

Contact