Skip to content
Back to skills

Linkedin Mcp

ASecurity

Use LinkedIn data for profile, company, job, post, feed, inbox, recruiting, and outreach work through the bundled LinkedIn MCP server. Use only for an explicit LinkedIn request.

  • 3,764 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 20, 2026
ai-agents

Works with

  • mcp

Security analysis

A100/100

Scanned October 4, 2026

npx -y skills add stickerdaniel/linkedin-mcp-server --skill linkedin-mcp --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Linkedin Mcp?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Linkedin Mcp
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/stickerdaniel-linkedin-mcp/badge)](https://www.skillsdirectory.com/skills/stickerdaniel-linkedin-mcp)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: linkedin-mcp
description: Use LinkedIn data for profile, company, job, post, feed, inbox, recruiting, and outreach work through the bundled LinkedIn MCP server. Use only for an explicit LinkedIn request.
---

# LinkedIn MCP

Use the bundled `linkedin` MCP server only when the user explicitly asks for
LinkedIn data or an action on LinkedIn. Do not call LinkedIn tools for unrelated
work, health checks, or background maintenance.

## Operating rules

- Start with the smallest read operation that can answer the request.
- Treat profile, company, job, post, feed, and message results as live LinkedIn
  evidence. Distinguish retrieved facts from inference.
- Keep searches and result pages modest. Do not collect data in bulk or spam.
- Never enable the plugin or its MCP server, edit Codex configuration, or start
  a login flow merely because LinkedIn might be useful. If either component is
  disabled, explain that state and stop.
- `send_message` and `connect_with_person` are write actions. Use them only when
  the user explicitly authorizes the exact recipient and action. Confirm the
  final message or connection note unless the user has already supplied it.
- Do not retry a failed write action unless the result proves it was not sent.

## Browser and authentication

The server uses a managed Chromium browser and the user's own LinkedIn session.
Installing or enabling the plugin does not sign in. The first LinkedIn data
request may prepare the browser, import an existing local session, or require a
visible login window. If LinkedIn presents a captcha, two-factor prompt, or
other user-owned authentication step, stop and ask the user to complete it.

If browser setup or authentication is still in progress, report that exact
state and retry once after it finishes. Do not loop, launch extra browser
instances, clear profiles, or replace the user's browser session.

## Tool selection

- Profiles: `get_my_profile`, `get_person_profile`, `search_people`, and
  `get_sidebar_profiles`.
- Companies: `get_company_profile`, `get_company_posts`, `search_companies`,
  and `get_company_employees`.
- Jobs: `search_jobs`, `get_saved_jobs`, `get_job_details`, and
  `get_job_apply_url`.
- Content: `get_feed` and `search_posts`.
- Messages: `get_inbox`, `get_conversation`, and `search_conversations`.
- Writes: `send_message` and `connect_with_person`, subject to the explicit
  authorization rules above.
- Cleanup: use `close_session` only when the user asks to end the managed
  browser session or when the current LinkedIn task is finished and no follow-up
  call is expected.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.