8 Tips to Create AI Skills: How They Differ from Prompts

Daniel - Multidisciplinary Writer on Finance, Business & AI
20 Min Read

Ever written a long prompt for an AI, got exactly the result you wanted, and then had to explain everything from scratch again the following week? Which sources to use, the writing style, the table layout, even the presentation colors, all typed out once more. It feels like having an assistant who forgets how we work every Monday, hehe.

That’s where learning to create AI skills really helps. A prompt is the request we give an AI in the moment. A skill is a package of how we work (instructions, examples, and reference material) that’s saved once and used again and again. So the difference isn’t about how long the text is, but about how the instructions are stored, called up, and used together with their supporting material.

It’s a bit like ordering at your favorite restaurant. The skill is the kitchen’s recipe and standards: the seasoning measurements, the cooking method, all the way to how the dish is plated. The prompt is today’s order: which dish, how many portions, and whether we’d like it less spicy. A saved recipe clearly makes life easier, but the kitchen still needs to know what we’re ordering.

In the AI series on danieel.id, skills are the second rung of the ladder: effective AI prompting makes a single request clear, a skill stores the rules you repeat every time, a simple AI workflow chains several steps together, and an AI agent plans its own steps. In my experience, skills are most useful for work whose technical requirements stay the same even when the material changes.

Worth noting: skill features in every AI service are still changing and can differ from one account to another. So we don’t stop at theory, I’ve put together eight tips to create AI skills below, each with an example skill and the prompt that calls it.

Start creating AI skills from one recurring need

How to create AI skills, starting from one recurring need such as product descriptions for a bakery

To create AI skills, it’s best to start with one task that you repeat often and that has a clear standard for the result, not with the wish that AI will help with everything at once. Take a bakery that adds new products to its catalog every week: the bread changes, but the way the descriptions are written stays roughly the same.

First, note down three things: the material you provide (input), what the AI does with it (process), and the result you want (output). A skill like “help with all my work” is too broad to evaluate.

Example skill — bakery-catalog

  • Task: write product descriptions of breads for the shop’s catalog.
  • Input: bread name, ingredients, weight, storage instructions, and price.
  • Technical requirements: 60–90-word descriptions written from a “we” point of view; prices formatted as Rp35.000 (Indonesian rupiah); allergens only from the owner’s statement.
  • Output: a title, a description, a Weight | Main ingredients | Allergens | Storage table, and a “Data not yet available” list, as an HTML draft.
  • Limits: don’t invent shelf life, certifications, or health claims.

Example prompt: “Use the bakery-catalog skill for this banana bread. Weight 250 grams, price Rp35.000. The ingredients and storage instructions are in the attachment.”

The prompt only carries today’s product data; the word count, price format, and table are already stored in the skill. For questions you rarely repeat, a regular prompt is enough.

Separate fixed rules from the needs of each request

A fixed instruction book and a new request card showing how skill rules are kept separate from daily prompts

When you create AI skills, each skill only needs to hold the rules that are always the same, such as language, tone, format, length limits, and what must be checked. Things that change, like an event date, a product name, or this week’s target, are better passed in through the prompt.

If the event date gets saved in the skill, next month’s event might end up with last month’s date, hehe. So creating AI skills also trains us to tell the difference between the work standard and the work content.

Example skill — community-caption

  • Task: write social media captions for neighborhood community activities.
  • Technical requirements: a friendly tone with no sarcasm; structure of opening → benefit → event details → call to action; a maximum of 80 words and two emojis; no text-speak abbreviations; two to three hashtags.
  • Output: three labeled caption options, each with its word count.
  • Limits: don’t guess a location or time that wasn’t mentioned.

Example prompt: “Use the community-caption skill for this Sunday’s river clean-up, 7:00 a.m. at the village bridge. Ask residents to bring gloves. Just two options this time.”

The prompt only needs to carry the new event details and change the number of options from three to two.

Lay out a workflow and quality criteria you can check

A university student mapping a flow from reading material to a summary and practice questions

When you create AI skills, good instructions take the form of a clear sequence of work, for example read the material, pick the important information, put the result together, then check that it fits. A line like “Be an amazing tutor” may sound cool, but it doesn’t explain what actually needs to be done.

Example skill — study-buddy

  • Task: help understand one chapter of course material and practice questions; the chapter changes every week, the way of studying stays the same.
  • Workflow: (1) read the whole material as the main source; (2) summarize it in 300–400 words; (3) explain 5–10 key terms; (4) give one analogy; (5) write five questions of increasing difficulty; (6) keep formulas exactly as they appear in the material.
  • Output: a summary, a Term | Meaning table, one analogy, and five questions without an answer key.
  • Limits: explanations of the answers are only given after the user sends their answers.

Example prompt: “Use the study-buddy skill for the attached Chapter 3. I already understand the definitions, so focus the questions on applying the concepts.”

The result is easy to check: the summary matches the material, the formulas stay correct, and the answer key doesn’t leak. What we get is a consistent process, not a guarantee that the AI is always right.

Add targeted examples, references, and limits

A creative worker comparing sample results with sketches and references as a guide for a skill

Sample writing, a glossary, and rules about what may be changed make an AI skill far more focused than adjectives like “professional.” Also explain what should be copied from the examples: the sentence style, not the content.

Example skill — story-editor

  • Task: edit the user’s story manuscript so it reads well, not write a new story.
  • Required references: a style guide, two sample paragraphs, and a glossary of preferred terms (for example, the Indonesian word “gawai” instead of the borrowed “gadget”).
  • Technical requirements: a maximum of four sentences per paragraph; fix spelling according to the official Indonesian spelling guide (EYD); keep the point of view, character names, quotes, and the ending.
  • Output: a clean manuscript, a Before | After | Reason table, and a list of things to confirm.
  • Limits: don’t add new experiences, characters, or events.

Example prompt: “Use the story-editor skill for the attached story. The readers are young parents; make the dialogue sound more natural.”

According to the Agent Skills specification, a skill is a folder containing a SKILL.md file with a name, a description, and instructions, plus references, templates, or scripts when needed. The structure looks roughly like this:

story-editor/
├── SKILL.md
└── references/
    ├── style-guide.md
    ├── sample-paragraphs.md
    └── glossary.md
---
name: story-editor
description: Edit the user's manuscript following the style guide; use for editing, not for writing new stories.
---
Read references/style-guide.md and references/glossary.md before editing.

Take the description seriously, because the AI uses it to decide when to call the skill.

Adapt the setup for ChatGPT, Gemini, and Claude

An instruction package applied through different features in ChatGPT, Gemini, and Claude

An AI skill design can be the same in ChatGPT, Gemini, and Claude, but the way you create and call it follows each service’s features. Mentioning a skill’s name in a prompt doesn’t automatically give the AI that skill.

The table below compares the three services in terms of how to create and call a skill and what to watch out for, according to their official guides.

ServiceHow to create and call itWhat to watch out for
ChatGPTCreate with @skill-creator; call with @Can be picked automatically when a task matches
GeminiSettings → Skills; call with /Gems on personal accounts are becoming skills
ClaudeUpload a ZIP via Customize → SkillsCode execution must be turned on

In other words, the package’s contents can travel anywhere, but how you install it needs adjusting. OpenAI’s official guide treats instruction-only skills as the starting point. Google’s help page explains that Gems on personal Google accounts transition to skills starting in November 2026, while work and school accounts follow in 2027. Claude’s Skills guide requires code execution to be turned on.

Example skill — meeting-summary

  • Task: turn meeting notes into a summary and a list of follow-ups.
  • Technical requirements: separate information, proposals, and decisions; a summary of no more than 150 words; missing data written as “not yet decided.”
  • Output: a summary, a list of decisions, and a Task | Owner | Deadline table.
  • Limits: don’t turn opinions into decisions, and don’t send the draft to attendees.

Example prompt: “Use the meeting-summary skill for the attached marketing meeting notes. Highlight tasks that don’t have an owner yet.”

Read also:

ChatGPT vs Gemini vs Claude: Which One Fits Your Needs?

What Is an AI Agent? How It Differs from Chatbots, Skills, and Workflows (with Real Examples)

Combine skills with prompts that stay specific

An AI assistant combining skill instructions with the specific needs of the latest prompt

Combining a skill with a prompt is most useful for work with many technical requirements: the skill stores all the standard rules, while a short prompt names today’s data and focus. A monthly report presentation with the same format every time is a good example. Pretty tiring to type it all out again and again, right?

Example skill — monthly-report-presentation (a design for a fictional business)

  • Source: the approved spreadsheet is the main source of figures; if it differs from operational notes, ask for confirmation.
  • Processing: calculate month-on-month changes only when the data is comparable; separate facts, suspected causes, and proposals.
  • Format: eight slides, no more than four points per slide; rupiah amounts in millions with one decimal place (written Rp125,4 juta in Indonesian format).
  • Design: the assets/template-laporan.pptx template; a navy #17324D and green #2A9D8F palette; line charts for trends, bar charts for comparisons.
  • Checks: match figures against the source; if the template isn’t available, say so plainly; don’t send the presentation to anyone.

Example prompt: “Use the monthly-report-presentation skill for the attached September data. The audience is the board of directors; highlight the sales drop in the eastern region.”

Next month, we only need to swap the data and the focus. Make sure the template file is included. When you try this out, use dummy data, not your company’s real figures. For financial reports, understanding the types of financial statements still helps us judge whether the analysis makes sense.

Test with normal cases, missing material, and different cases

A teacher reviewing test results from complete material, incomplete material, and a different case

Once you create AI skills, test them with three kinds of cases: complete material, missing material, and a request outside the skill’s scope. What gets judged isn’t just whether the result reads well, but also factual accuracy and whether the rules were followed.

Example skill — product-description

  • Task: write online store product descriptions from the business owner’s official specifications.
  • Technical requirements: a title of no more than 70 characters; an 80–120-word description; three benefits supported by the specifications.
  • Limits: don’t invent durability, certifications, or testimonials; don’t guess sizes.

Example test prompt: “Use the product-description skill for this cotton tote bag. The size and care instructions aren’t available yet. Write the description without guessing that information.”

If the AI suddenly writes “fits a 15-inch laptop” when the size is still unknown, the result fails no matter how appealing the language is. Recurring mistakes are fixed in the skill, missing data is filled in through the prompt. Change one rule at a time, then test again.

Maintain skills, protect data, and keep a human check

A user updating skill instructions while protecting confidential documents

After you create AI skills, maintain them like any work document: give them a version number, an update date, and a change log, then update them whenever the format or template changes.

Example skill — service-report

  • Task: turn anonymized customer complaint notes into a weekly report.
  • Technical requirements: group complaints by theme, sort them from the most cases, and separate complaints from verified findings.
  • Output: a summary of no more than 100 words and a Theme | Number of cases | Verification status | Proposal table, labeled “DRAFT – for review.”
  • Limits: don’t copy phone numbers, addresses, or customer identities; don’t publish the report.

Example prompt: “Use the service-report skill for this week’s anonymized notes. Focus on delivery delays; prepare a draft for Friday’s meeting.”

For skills you share, use dummy data and never store passwords or OTPs in them. Read through the instructions and scripts of skills made by others first, because the scripts inside get run by the AI too. The habit of securing your digital accounts still matters.

Finally, check and double-check. A skill keeps the standard, but it doesn’t guarantee the AI is always right, so we still filter facts and figures ourselves, just as we do when facing online hoaxes.

Frequently Asked Questions About AI Skills

What is the difference between an AI skill and a prompt?

An AI skill is a package of instructions, examples, and reference material saved for repeated use, while a prompt is a request for what you need in the moment. The skill stores the format and limits that never change, while the prompt carries today’s data.

Do you need coding skills to create AI skills?

You don’t need coding skills to create AI skills, because the simplest skill is just a SKILL.md file with a name, a description, and instructions in everyday language. Scripts are only needed for technical tasks such as processing data.

Can one skill be used in ChatGPT, Gemini, and Claude?

The contents of one skill can be reused in ChatGPT, Gemini, and Claude because the SKILL.md folder format is open. However, the way you upload it, call it, and the tools it supports differ, so the skill needs retesting after you move it.

How long should a SKILL.md file be?

The Agent Skills specification recommends keeping the main body of SKILL.md under 500 lines. Long examples and style guides are better stored in the references folder, so the AI reads them only when needed.

What is the difference between an AI skill, an AI workflow, and an AI agent?

An AI skill stores the standard way to do one type of task, an AI workflow chains several steps in a fixed order, and an AI agent plans its own steps based on a goal. Workflows and agents work more neatly when each of their steps follows a clear skill.

Closing: Save the Recipe Once, Order What You Need

A student, a creative worker, and a business owner reviewing AI results together before using them

In short, to create AI skills means picking one recurring task, saving its fixed rules along with examples and limits, calling it with a specific prompt, and then testing and maintaining it.

Just like that favorite restaurant, the recipe only needs to be saved once, while the order can change every day. Try starting this week: note down the rules you retype most often, then turn them into your first skill. Who knows, next month we might thank ourselves for not having to explain the same thing again.

Of all your routine tasks, which one most deserves to be your first skill? Share it in the comments; your idea might help other readers.

Hope this helps.

This post is also available in: Indonesia

Share This Article
Multidisciplinary Writer on Finance, Business & AI
Follow:
I'm Daniel, B.Eng., MBA, born and raised in Sumatra, Indonesia, with a background in chemical engineering and a Master of Business Administration (MBA). On danieel.id, I write about finance and company analysis, the practical use of AI for learning and work, and personal notes on Islam, science, history, health, and everyday life. Health articles here are for general education, drawn from trusted sources, and are not a substitute for professional medical advice.
Leave a Comment