Skip to content

CLI reference

LinkedIn MCP Pro ships with 10 CLI commands for headless operation.

Quick reference

Command Purpose
linkedin-mcp-pro Run the stdio MCP server (for AI agents)
linkedin-mcp-web Start the web dashboard
linkedin-mcp-install Wire an MCP host + manage accounts
linkedin-mcp-login Headless cookie import
linkedin-mcp-templates Manage post templates
linkedin-mcp-schedule Manage scheduled posts
linkedin-mcp-analytics Export analytics (CSV / JSON)
linkedin-mcp-stats Show account statistics
linkedin-mcp-health Diagnostics + readiness check
linkedin-mcp-deadman Watchdog for stalled automations

linkedin-mcp-pro

Run the stdio MCP server. This is what AI agents launch.

linkedin-mcp-pro serve

Options:

Flag Default Description
--transport stdio stdio (default) or sse
--port 8081 Port for SSE transport

Exit codes:

  • 0 — clean shutdown
  • 1 — fatal error (see logs)
  • 130 — interrupted (Ctrl+C)

linkedin-mcp-web

Start the dashboard.

linkedin-mcp-web [--host HOST] [--port PORT] [--reload]
Flag Default Env var Description
--host 127.0.0.1 LINKEDIN_MCP_WEB_HOST Bind host
--port 8080 LINKEDIN_MCP_WEB_PORT Bind port
--reload off — Auto-reload on file changes (dev)

Examples:

# Local only
linkedin-mcp-web

# LAN-accessible
linkedin-mcp-web --host 0.0.0.0 --port 8080

# Dev mode with auto-reload
linkedin-mcp-web --reload

linkedin-mcp-install

Wire an MCP host (Claude Desktop, Cursor, etc.) + manage accounts.

Subcommands

add <agent>

Add LinkedIn MCP Pro to an MCP host.

linkedin-mcp-install add claude-desktop
linkedin-mcp-install add cursor
linkedin-mcp-install add cline
linkedin-mcp-install add windsurf
linkedin-mcp-install add zed
linkedin-mcp-install add continue
linkedin-mcp-install add open-webui
linkedin-mcp-install add librechat
linkedin-mcp-install add lobechat

Supported agents: claude-desktop-mac, claude-desktop-linux, claude-desktop-win, claude-code, cursor, cline, continue, windsurf, zed, open-webui, librechat, lobechat.

verify <agent>

Check if a host is correctly wired.

linkedin-mcp-install verify claude-desktop

Print the JSON snippet for every supported agent (useful for manual setup).

linkedin-mcp-install print-configs

account <action>

Manage accounts.

linkedin-mcp-install account add
linkedin-mcp-install account list
linkedin-mcp-install account default <name>
linkedin-mcp-install account pause <name>
linkedin-mcp-install account resume <name>
linkedin-mcp-install account remove <name>

linkedin-mcp-login

Headless cookie import.

linkedin-mcp-login                # interactive
linkedin-mcp-login --li-at X      # paste cookie value
linkedin-mcp-login --account X    # save to specific account

linkedin-mcp-templates

linkedin-mcp-templates list
linkedin-mcp-templates add --name "Hiring post" --body "..."
linkedin-mcp-templates update --id X --body "..."
linkedin-mcp-templates delete --id X
linkedin-mcp-templates export --format json > templates.json
linkedin-mcp-templates import templates.json

linkedin-mcp-schedule

linkedin-mcp-schedule add --draft-id X --at "2026-01-15T14:00:00Z"
linkedin-mcp-schedule list
linkedin-mcp-schedule cancel --id X
linkedin-mcp-schedule worker                       # foreground worker
linkedin-mcp-schedule worker --detach              # background worker

linkedin-mcp-analytics

linkedin-mcp-analytics summary --days 30
linkedin-mcp-analytics posts --format csv > posts.csv
linkedin-mcp-analytics engagement --format json
linkedin-mcp-analytics best-times                 # shows heatmap

linkedin-mcp-stats

Show account statistics (followers, engagement, top posts).

linkedin-mcp-stats
linkedin-mcp-stats --days 7
linkedin-mcp-stats --format json

linkedin-mcp-health

Diagnostics + readiness check.

linkedin-mcp-health

Output:

linkedin-mcp-pro v2.3.1 — ready
Python 3.11+
DB: ~/.linkedin-mcp/profile/state.db
Browser: available
Cookies: present (default account, expires in 87 days)
LLM keys: 3 configured
Scheduler worker: running (PID 12345)

Exit codes:

  • 0 — healthy
  • 1 — warning (some subsystem degraded)
  • 2 — error (something broken)

linkedin-mcp-deadman

Watchdog for stalled automations.

linkedin-mcp-deadman check                     # one-shot check
linkedin-mcp-deadman watch                     # foreground watch
linkedin-mcp-deadman watch --interval 5m       # custom interval

Alerts via:

  • Console (always)
  • Telegram (if TELEGRAM_BOT_TOKEN + TELEGRAM_CHAT_ID set)
  • Webhook (if LINKEDIN_MCP_DEADMAN_WEBHOOK set)

Triggers:

  • No posts fired in 24h (when scheduled posts exist)
  • No comments/likes in 24h (when engagement enabled)
  • Scheduler worker not running
  • Cookies expired or invalid