How to Use Claude for Social Media Research and Analysis (2026)

Published July 2026.

You can now hook up Claude Code, or any AI agent, to SocialKit and have it pull real social media data for you. Transcripts, comments, stats, and plenty more, from any YouTube, TikTok, Instagram, Facebook, LinkedIn, or X URL. You ask a question, it goes and gets the data, then works with it. No scrapers, no proxies, no wrapper code.

Here's how to set it up.

Step 1: Go to socialkit.dev/ai-agents

Head to socialkit.dev/ai-agents. That's where you connect SocialKit to your AI tools, either through MCP or Claude Code Skills. The same access key works for both.

Option A: MCP (Claude Desktop, Cursor, and more)

Copy the MCP config from the page.

Drop it into your claude_desktop_config.json or mcp.json:

{
  "mcpServers": {
    "socialkit": {
      "url": "https://mcp.socialkit.dev/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

Paste it into Claude and tell it to connect. It confirms, and you're done.

Now you can say things like:

• "Show me MrBeast's last 10 videos and their stats"
• "Summarize this TikTok and tell me if the top comments are positive or negative"
• "Pull the transcript of this Instagram Reel and list the hooks"
• "What are people actually saying in the comments of this LinkedIn post?"

Claude figures out which SocialKit tools to call, and in what order, then hands you the answer.

Option B: Claude Code Skills

If you live in the terminal with Claude Code, install the SocialKit skills instead:

npx skills add SocialKit-API/skills --all

Skills repo: github.com/SocialKit-API/skills

This gives Claude Code direct access to SocialKit actions as terminal skills. You can chain them into whatever else you're already doing, like dropping a batch of transcripts straight into your own analysis script.

It works on every platform, not just YouTube

The same setup covers all of them: YouTube, TikTok, Instagram, Facebook, LinkedIn, and X/Twitter. Transcripts, comments, stats, profiles, posts, downloads, search. One key, one conversation. Paste a URL from any of them and ask.

What people actually do with it

Content research. Ask what's working in your niche, pull the transcripts of the top videos, and see the hooks and structure they use before you write your own.

Competitor tracking. Grab a competitor's recent posts and their view counts in one shot, and spot which videos are actually carrying the account.

Audience research. Fetch the comments on a post and let Claude tell you what people think, instead of scrolling three hundred replies yourself.

Repurposing. Transcript to summary to a first draft, all in one conversation, whatever platform the original came from.

That's it

A chat window that reaches straight into social media data. No dashboards, no CSV exports, no glue code. SocialKit deals with the platform APIs and the anti-bot mess, Claude does the thinking.

Get started at socialkit.dev/ai-agents.