AI goes beyond MCP

On 17 Oct 2025, Anthropic announced Claude Skills. Initially I thought that it will probably be just another small upgrade to their AI Models, but little did I know it would be a feature that may rival [MCP](https

//modelcontextprotocol.io/docs/getting-started/intro).

Benefits of Claude Skills

Claude Skills provide workflows, resources and examples to the model which allows it to generate and create more useful content that can be integrated into workflows. Another key benefit is that it is very easy to customize and alter. Most scripts are written in Python and are pretty easy for new developers to understand. Even if you don’t know programming, simple skills can still be created using just [Markdown](https://en.wikipedia.org/wiki/Markdow n).

Types of Skills

There are basically 3 types of skills - Metadata, Instructions, and Resources & Code.

Metadata

These Skills are always loaded at startup and only have a name and description from YAML frontmatter. They are quite token efficient and use approximately 100 tokens per skill.

---
name: PDF Processing
description: Extract text and tables from PDF files, fill forms, merge documents. Use when working with PDF files or when the user mentions PDFs, forms, or document extraction.
---

Instructions

These Skills are loaded when triggered and contain a SKILL.md body with instructions and guidance. They usually use less than 5,000 tokens.

# PDF Processing

## Quick start

Use pdfplumber to extract text from PDFs:

import pdfplumber

with pdfplumber.open("document.pdf") as pdf:
text = pdf.pages[0].extract_text()

For advanced form filling, see [FORMS.md](FORMS.md).

Resources & Code

This is the most advanced type of skill. It contains examples, scripts and templates. This is an example of such a skill’s directory:

pdf (skill name)
├── LICENSE.txt
├── SKILL.md
├── FORMS.md
├── REFERENCE.md (detailed API reference)
├── scripts
│   ├── check_bounding_boxes.py
│   ├── extract_form_field_info.py
│   ├── fill_fillable_fields.py
│   └── fill_pdf_form_with_annotations.py
└── templates
├── academic_paper.py
└── portfolio.py

• SKILL.md - Contains generic text and description, similar to a Readme,
• FORMS.md - Form-filling guide,
• REFERENCE.md - Detailed API reference,
• scripts - This directory has scripts and workflows that the model can access,
• templates - This directory contains all the resources and templates the model can use.

Skills vs MCP

Anthropic released MCP last year as an open standard, open-source framework to standardize the way artificial intelligence (AI) systems like large language models (LLMs) integrate and share data with external tools, systems, and data sources.

While both Skills and MCP are similar, they both also have their own differences. MCP connects AI to external services and data sources, Skills gives the AI resources to work with. These resources can teach the AI model how to use tools. Simply put, MCP provides tools, Skills teaches the AI to use them.

Writing a SKILL.md

Here is a SKILL.md template provided by Anthropic:

---
name: Your Skill Name
description: Brief description of what this Skill does and when to use it
---

# Your Skill Name

## Instructions
[Clear, step-by-step guidance for Claude to follow]

## Examples
[Concrete examples of using this Skill]

The only required fields are the name and description. The name field has a 64 characters limit while the description field has a 1024 character limit.

Conclusion

Anthropic Skills is gonna be a real game-changer, with AI capable of running workflows and scripts. I’m really looking forward to it being integrated into other systems.

More Info & References

Simon Willison’s Weblog - https://simonwillison.net/2025/Oct/16/claude-skills

Claude Support - https://support.claude.com/en/articles/12512176-what-are-ski lls

Claude Docs - https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview#level- 2%3A-instructions-loaded-when-triggered