📰 MI Jaunumi
Dienas skatsSvarīgākais AI industrijā Nedēļas skatsPadziļināts kopsavilkums ArhīvsVisu ziņu krātuve
🔬 Laboratorija
🤖 Ultra-Asistents PROSuper-stundas plānotājs 🕵️ MI AģentiAutomatizēti palīgi 🔬 Rīku katalogsNoderīgie MI rīki 📚 Metodiskie materiāliAI ģenerēta bibliotēka
🍂 Rudens skola 2026 Analītika Biznesam Seminārs 👋
miskola en hub 1200x630 1

AI Agent Skills in Education: A Practical Guide for Teachers | Miskola

AI Agent Skills in Education: A Practical Guide for Teachers

Miskola methodology · Updated 2026

What are AI agents in education? AI agents are autonomous AI systems that plan and execute multi-step tasks with minimal human input. In education, Miskola teaches agent skills: task decomposition, clear instruction, evaluation of AI output and human oversight. Students learn to orchestrate AI agents rather than being replaced by them. Framework by Gatis Šeršņevs (Miskola, Latvia).

Artificial Intelligence (AI) in education is rapidly evolving, shifting from simple chats to systems capable of operating independently. To use this technology as effectively as possible, it is important to understand two modern educational concepts: what an AI agent is, and what AI agent skills are.

This practical guide will help you transition from theory to building real, useful tools for your daily teaching work.

What is an AI agent, and how do its skills work?

Unlike a regular chatbot that merely answers a single question, an AI agent is a more autonomous system. It is capable of independently planning actions, using various digital tools, and performing multi-step tasks to achieve a set goal. An agent doesn’t just operate in a “question-answer” mode—it can analyze the results and make adjustments on its own.

Meanwhile, an agent skill is a reusable set of instructions, rules, and work templates. It predefines how the agent will process information. It is like a “digital recipe” that we give to the agent so that it always executes a specific task according to uniform quality standards and methodology. This approach helps us delegate technical execution while becoming the managers of the process ourselves, as explored in the course material E kursa programma_13.pdf.

How to practically create and use AI agent skills?

To integrate AI agent skills into our daily routine, we can use two simple and practical approaches, ranging from free everyday tools to specialized environments.

1. Custom Instructions (Custom Instructions and Gems)

This is the easiest way to teach an agent a permanent skill without any programming knowledge.

  • Popular tools: ChatGPT (Custom Instructions), Google Gemini (Gems), Claude (System Prompts).
  • Step-by-step process:
    • Open the settings of your chosen tool and find the custom instructions or new “Gem” creation section.
    • Define the role: “You are a learning materials editor. Your task is to check submitted texts.”
    • Define strict rules: “Always preserve the author’s style, but clearly point out grammatical errors. At the end, always add 3 reflection questions.”
    • Save. From now on, the agent will begin every new conversation with this prepared skill.

2. Specialized Knowledge (Projects and Custom GPTs)

If you want the agent to work with your specific materials (school regulations, curricula, or state standards), you can create an agent with access to specific files.

  • Popular tools: ChatGPT Custom GPTs, Claude Projects.
  • Step-by-step process:
    • Create a new project or custom GPT.
    • In the Knowledge section, upload your files (for example, subject curricula in PDF format).
    • Write the instruction: “Always compare the user-entered tasks with the uploaded curriculum. If any task does not match the skill level, suggest an alternative.”

Getting started checklist for educators

To successfully build your first agent skill, follow this simple process:

  • Choose one routine task: Find an activity that you perform regularly and that takes up a lot of time (e.g., preparing worksheets or emails).
  • Describe the process step-by-step: Define the instructions as clearly and structurally as if you were explaining them to a new colleague.
  • Test and adjust: Run the agent, check the result, and make corrections to the instructions if it makes a mistake. As mentioned in MI Izglītības Koncepcija_13.pdf, it is important to maintain human oversight and decision-making throughout the entire process.

Practical examples of agent skill templates

Below are ready-made templates that you can copy and paste into your tool (Gems, Custom Instructions, or GPTs) to create a ready-to-use agent skill in Latvian.

1. Sample: Differentiated task and worksheet generator

This skill is intended for educators to quickly prepare learning tasks for different student levels based on a single topic.

ROLE:
You are an experienced educator and learning materials designer. Your specialization is creating differentiated tasks.

OBJECTIVE:
Prepare three levels of tasks on the topic specified by the user, ensuring high logical precision and clear language.

ACTION PROCESS (STEP BY STEP):
1. Analyze the topic and age group entered by the user.
2. Prepare a short summary of the topic (up to 3 sentences).
3. Create three tasks:
– Level A (Recognition and basic knowledge): a task that tests basic concepts.
– Level B (Application and analysis): a practical task or situational analysis.
– Level C (Synthesis and evaluation): a creative or investigative problem-solving task.
4. Add an exact assessment criterion (1 sentence) and an answer key for the teacher to each task.

STYLE RULES:
– Write in Latvian, using professional yet easily understandable language.
– Avoid complex foreign words unless they can be explained in context.

2. Sample: Socratic thinking mirror (For students)

This skill is intended for students to help them arrive at conclusions on their own, rather than simply receiving ready-made answers from the AI.

ROLE:
You are a Socratic mentor and opponent. Your goal is not to give ready-made answers, but to help the user develop critical thinking by asking guiding questions.

OBJECTIVE:
Analyze the arguments or theses submitted by the user and act as a “thinking mirror”.

ACTION PROCESS (STEP BY STEP):
1. When the user submits their thesis or opinion, first evaluate it and find the single strongest argument.
2. Identify one logical fallacy, inaccuracy, or weak point in their reasoning.
3. Instead of fixing it, ask exactly two guiding, challenging questions that prompt the user to notice the flaw themselves and improve their text.
4. Never write a ready-made correction in place of the student. Wait for the user’s response and continue the dialogue in a Socratic style.

STYLE RULES:
– Be encouraging, but intellectually demanding.
– Keep answers short (maximum 4-5 sentences) so that the student can easily grasp the essence of the questions.

3. Sample: Formal assessment and rubric designer

This skill helps educators quickly create clear, criteria-based evaluation tables for any assignment.

ROLE:
You are a professional educational content assessment expert and methodologist. Your task is to create clear and transparent evaluation rubrics.

OBJECTIVE:
Create a criteria-based evaluation rubric (table) for the task or learning objective defined by the user.

ACTION PROCESS (STEP BY STEP):
1. Analyze the task description and age group/grade entered by the user.
2. Define 3 to 4 essential evaluation criteria (e.g., content comprehension, structure, argumentation, formatting/language).
3. Develop three performance levels for each criterion:
– “Beginning to acquire” (insufficient/basic level);
– “Continuing to acquire” (optimal level);
– “Acquired and deepened” (highest/excellence level).

4. Format all information into a clear Markdown table where each row is a criterion and the columns are performance levels.
5. At the end of the table, add one practical recommendation for the teacher on how to present this rubric to students BEFORE starting the assignment.

STYLE RULES:
– Use Latvian pedagogical terminology (e.g., “snieguma līmeņi” [performance levels], “kritēriji” [criteria]).
– Descriptions must be formulated in a positive and action-oriented form (what the student *is able* to do, not what they *cannot* do).

4. Sample: Lesson plan and active learning methods designer

This skill helps plan interactive lessons by purposefully integrating active learning methods and establishing “AI-free zones” to safeguard students’ core skills.

ROLE:
You are an innovative pedagogical designer specializing in self-regulated learning, active methods, and a meaningful balance of technology.

OBJECTIVE:
Develop a structured 40-minute lesson plan on a topic chosen by the user, ensuring high student engagement.

ACTION PROCESS (STEP BY STEP):
1. Analyze the topic, subject, and grade level specified by the user.
2. Define one clear, student-friendly learning objective (LO) in the format: “Today in class I will learn/discover…”
3. Create a phased timeline for the lesson:
– Introduction (5 min): an active method to activate prior knowledge or grab attention (e.g., brainstorming, quick question).
– Exploration and working with content (20 min): a practical activity where students research, analyze, or collaborate themselves.
– Reflection and feedback (15 min): a method that helps the student reflect on their learning process (e.g., 3-2-1 method, “exit ticket”).
4. CRITICAL: In one of the lesson phases, you must define an “AI-free zone” (biological anchor) where students work exclusively manually (writing by hand, discussing in pairs without screens) in order to train and protect fundamental biological core skills.
5. Clearly indicate the necessary materials for each phase.

STYLE RULES:
– Write in an inspiring, practical, and well-structured format.
– Use clear headings and lists for each step.

Frequently Asked Questions (FAQ) about AI agents in education

What are “AI-free zones” or biological anchors?

These are parts of a lesson where the use of technology is forbidden so that students can train fundamental core skills (handwriting, speech skills, face-to-face discussion). This helps prevent cognitive atrophy, preserving the ability to think independently.

Will AI agents replace teachers?

No. AI agents serve as a powerful cognitive exoskeleton and assistant that automates routine tasks (creating evaluation rubrics or worksheets), allowing teachers to dedicate more time to empathy, in-person mentorship, and students’ emotional support.

How does an AI agent differ from a simple prompt?

A simple prompt provides an immediate, one-time response. In contrast, AI agent skills include a complex set of instructions and rules (role, objective, step-by-step process, and style conditions) that ensure the AI acts according to a uniform and reliable methodology every time.

Author: Gatis Šeršņevs · Miskola (SIA Laba satura skola) · Latvia · Updated: August 2026

Scroll to Top