Claude Skills for Social Media Data: Adding Live Creator Data to Claude Code

Anthropic shipped Claude Skills in late 2025 and the format has quietly become the cleanest way to give Claude domain knowledge it does not already have. A skill is a folder. The folder has a SKILL.md file that tells Claude when to invoke the skill and how to use it. That is the whole format. No build step, no runtime, no server. You drop the folder into ~/.claude/skills/ and Claude reads it on demand.

This post explains what skills are, where they fit alongside MCP and tools, and how to install CreatorCrawl's skill for live social media data into Claude Code in about 30 seconds.

#What a Claude Skill actually is

A skill is a small Markdown file with frontmatter. The frontmatter has a name and a description. The description is the most important field. Claude reads the descriptions of every available skill and decides which ones (if any) match the user's request. If it picks yours, it reads the full body and follows the instructions inside.

A minimal skill looks like this:

---
name: my-skill
description: When the user wants to do X, run this skill. Triggers on phrases like "X" or "Y".
---

# My Skill

Step 1: do this.
Step 2: do that.

That is it. There is no Python, no JavaScript, no tool calls necessary. The skill body is a prompt that Claude follows when invoked.

#Where skills fit alongside MCP

MCP and skills sit in different layers. They are complementary, not competing.

• MCP gives Claude programmatic tools. The Claude runtime calls a function on an MCP server and gets structured data back.
• A skill gives Claude instructions on when and how to use those tools. The skill itself does not call anything. It teaches Claude the workflow.

A good skill for an MCP-backed product looks like a small playbook: "the user wants to find creators, here is which MCP tools to call, in what order, and how to format the results." The skill makes Claude better at orchestrating the tools the MCP server exposes.

For a product like CreatorCrawl that ships both an MCP server (with 60+ tools) and a skill (with the patterns for using them), the combination removes most of the prompt engineering an end user would otherwise have to do.

#What CreatorCrawl's skill does

The current CreatorCrawl skill covers the TikTok and Instagram workflows that show up repeatedly (YouTube, LinkedIn, Twitter, and Reddit are reachable through the MCP server today and will land in the skill next). The patterns it ships with:

• Find creators in a niche.
• Pull recent posts and rank them by engagement.
• Pull a TikTok or YouTube transcript and summarise.
• Run a competitor audit across two or more handles.
• Build a daily digest from a list of accounts.

When you ask Claude something that matches one of these patterns ("find me fitness influencers on Instagram"), the skill activates, the MCP tools get called in the right order, and Claude returns structured output. You did not have to know which tool to ask for. The skill handled that.

#Install in Claude Code

If you have Claude Code already, this takes 30 seconds.

mkdir -p ~/.claude/skills/creatorcrawl
curl -sSL https://creatorcrawl.com/skills/creatorcrawl/SKILL.md \
  -o ~/.claude/skills/creatorcrawl/SKILL.md

That is the whole install. Open any Claude Code session and try a prompt like:

"Find 5 TikTok creators in the cold brew coffee niche with between 50k and 500k followers. Pull their last 5 videos and rank by engagement."

Claude will see the skill description, route the request through it, and call the MCP tools in the right order. If the MCP server is also configured (see the next section), you get structured results back. If not, Claude tells you how to set it up.

#Pairing the skill with the MCP server

The skill is the playbook. The MCP server is the engine. You want both. If you have not added the MCP server yet:

{
  "mcpServers": {
    "creatorcrawl": {
      "url": "https://app.creatorcrawl.com/api/mcp",
      "headers": {
        "x-api-key": "your_api_key_here"
      }
    }
  }
}

Drop that in ~/.claude.json under the mcpServers key (or use the equivalent setting in Claude Desktop, Cursor, Windsurf).

With both installed, Claude has the data layer (MCP tools), the workflow patterns (skill), and the natural-language interface (the chat) covered. You write a prompt, Claude does the rest.

#Why ship a skill at all

Three reasons.

Skills shorten the prompt the user has to write. Without a skill, the user has to know that to find creators you call search_users, then get_profile for each result, then get_profile_videos, then compute engagement. With a skill, the user says "find me creators" and the skill handles the rest.

Skills are discoverable inside Claude Code. Claude reads every skill description at session start. A well-written description means the right skill activates on the right prompt without the user invoking it explicitly.

Skills are a distribution channel. A skill file is just text. Users can share it, fork it, modify it. We publish ours so anyone using Claude Code can pull it down and have CreatorCrawl available immediately.

#When to write your own skill instead of using ours

If you have a specific workflow that does not match the patterns CreatorCrawl ships, write a skill on top of the MCP tools. The MCP server exposes the raw capability, the skill captures your specific workflow. Both layers are open enough that you can build the part that matches your product.

#TLDR

A Claude Skill is a markdown file that teaches Claude when and how to use tools. CreatorCrawl ships one for social media data. Install it, pair it with our MCP server, and Claude Code can find creators, pull posts, run competitive audits, and write digests without you writing orchestration code.

Sign up for free for an API key. Read the MCP docs for the full tool list. Or curl the skill file straight into ~/.claude/skills/ and try a prompt.