What are Claude Skills? Use Skills to create PPT, Word and Excel files without installing Claude Code

Search interest in "Claude Skills" has grown quickly over the past year. Many people first hear about Skills in discussions about Claude Code, but Skills are not just for developers. They solve an everyday problem: how to make an AI follow your method every time, without explaining it again in each conversation.

What are Claude Skills?

Skills (officially called Agent Skills) were introduced by Anthropic in October 2025 as a way to extend what Claude can do. A Skill is a folder whose core is a SKILL.md file:

• A short metadata header gives the Skill a name and a description of when it should be used.
• The body explains, in plain language, the steps, requirements and checks for the task.
• The folder can also hold templates, reference material or scripts that the Skill uses.

Claude reads only each Skill's name and description up front, and loads the full content when a task calls for it. You can keep many Skills available without filling the conversation context.

Anthropic provides Skills for working with Word, PowerPoint, Excel and PDF files, which is why many people meet Skills when they use Claude to create office documents.

Skills vs. prompts vs. MCP

	Prompt	Skill	MCP
What it is	Instructions in one conversation	A packaged, reusable method	A connection to external tools and data
Best for	One-off tasks	Repeated tasks with stable steps	Reading or writing external systems such as databases or calendars
How you reuse it	Copy and paste each time	Save once, use when needed	Configure the connection once

In short: a prompt says how to do this task, a Skill says how to do this kind of task from now on, and MCP lets the AI use an outside system.

What a SKILL.md looks like

A simplified Skill for weekly reports:

---
name: weekly-report
description: Turns scattered work notes into a weekly report Word document in a fixed format. Use when the user mentions a weekly report or weekly summary.
---

# Weekly report

## Inputs
- This week's work notes (chat logs, to-do lists or rough notes)
- Next week's plan, if available

## Steps
1. Group the notes by project and merge duplicates.
2. For each project, write progress, results and issues. Quantify results where possible.
3. List issues that need a manager's help separately.

## Output
- A Word document titled "Weekly report (date range)"
- No longer than one page

## Checks
- Do not invent numbers that were not provided; mark missing information as [to be added]

A good Skill has a description that says when to use it, steps specific enough for someone else to follow, and a clear list of what a person must verify.

Using Skills without installing Claude Code

Running Claude Code locally means installing a command-line tool and setting up an environment, which is a hurdle if you don't write code. If your goal is to create and work with documents, you can use Skills in the browser instead. In Jinghe AI:

1. Choose a model from the model menu in the chat.
2. Use system Skills: the platform already includes Skills for PPT, Word and Excel files. Describe your task and start.
3. Save your own Skills:
   • In My Skills, choose Create and write the name, when to use it and the steps;
   • or choose Import, upload an existing Skill file (up to 10 MB) and review the preview before saving;
   • or ask in the chat to save the workflow you just discussed as a Skill, then confirm the preview.
4. Reuse it: select the Skill in a new task and the AI follows your saved method.

My Skills are private to you and sync between the web app and the desktop client.

What makes a good Skill

• Documents with a fixed format: weekly or monthly reports, meeting minutes, project updates.
• Analyses with fixed steps: monthly sales reviews, survey summaries, competitor comparisons.
• Team standards: company slide layouts, document formats, required report sections.

A simple rule: if you have explained the same task to an AI three times, it is worth turning into a Skill.

Things to keep in mind

• Read an imported Skill before saving it. Skills shape how the AI behaves, so make sure the steps are what you want.
• Verify important results. Skills make output more consistent, but facts, numbers and conclusions still need a human check.
• Check the interface for current features. Available models, Skills and membership permissions can change as the product evolves.