mcpField
L
linkedin-data

linkedin-data

Remote
superagnt.comSocial Media6Updated today

About linkedin-data

A superagnt data source for AI agents researching LinkedIn profiles, companies, jobs, posts, and search results. It supports sales research and recruiting within a broader MCP server and workflow ecosystem.

Key features

  1. 01LinkedIn profile and company data
  2. 02Job and post data
  3. 03LinkedIn search
  4. 04Sales and recruiting research
  5. 05Packaged agent workflows

Use cases

  1. 01Researching sales prospects
  2. 02Finding candidates and reviewing job information

MCP support

The superagnt MCP server exposes LinkedIn and other data sources to AI agents. Packaged skills provide workflows that use its tools.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.superagnt.com/mcp/linkedin
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

superagnt plugins for Claude Code

The superagnt MCP server gives your agent every data source (X, TikTok, YouTube, Instagram, Reddit, LinkedIn, web), people + company enrichment, a real workspace Postgres, schedules, webhooks and shareable dashboards. These plugins carry the skills that teach the workflows.

Install

  1. Add the superagnt server to Claude as a connector: open the prefilled dialog and click Add. The consent screen doubles as signup. One connector covers claude.ai, the desktop app, mobile, Cowork and Claude Code signed in with your Claude account. The desktop app only loads connectors when it starts, so refresh it with Cmd+R (Ctrl+R on Windows) or quit and reopen it after you approve.
  2. Install the skills:
/plugin marketplace add superagnt/plugins
/plugin install superagnt@superagnt

No tokens, no config editing, no secrets in this repo. Claude Code signed in with an API key doesn't load connectors; use the fallback in the setup guide, https://www.superagnt.com/install.

Skills

Skill plugins are packaged workflows: the skill plus the tools it needs. Each pulls the base plugin in as a dependency, so one line installs the whole stack:

/plugin install super-seo@superagnt             # your whole SEO practice, measured and ranked
/plugin install audience-radar@superagnt        # cited post ideas, every morning
/plugin install competitor-mindshare@superagnt  # a daily competitor scoreboard on X
/plugin install gtm-prospecting-desk@superagnt  # scored prospects + drafts, every morning

Every skill has a page describing exactly what it does before you install anything: https://superagnt.com/skills

What's inside

  • plugins/superagnt/ — eight base skills (data-sources, lead-generation, workspace-db, task-agents, automations, data-pipelines, dashboards, superagnt-setup) and the /superagnt:connect command (the Codex copy also carries the MCP server config).
  • plugins/<skill>/ — one skill per packaged workflow. Read any skill before installing; they are plain markdown.

Two rules every skill follows: your agent can propose a plan upgrade when a limit is reached, but money only ever moves on a confirmation page you open yourself, and nothing sends outbound (email, posts, DMs). Drafts are the deliverable.

Other harnesses

Not on Claude Code? The same setup is one paste in any MCP client. Open https://www.superagnt.com/install, copy the line it shows, and paste it into your agent.

About this repo

Generated from the superagnt monorepo (tools/claude-plugins/) — issues and PRs here are read, but content changes land upstream and sync out. MIT licensed.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.