> ## Content Index
> Fetch the complete content index at: https://blog.jestor.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# How to Write Instructions for an AI Agent
- URL: https://blog.jestor.com/write-instructions-for-an-ai-agent/
- Published: 2026-09-17T22:38:05.000Z
- Updated: 2026-09-17T22:38:05.000Z
- Author: Bruno Bannach

To **write instructions for an AI agent**, describe its role, its tone, and its limits. What it should know goes in the knowledge base; what it may change goes in field permissions.

## What belongs where

Instructions carry more than they should when the other two are neglected.

- **Instructions** — role, tone, how to handle uncertainty, when to escalate
- **Knowledge base** — the facts and policies it answers from
- **Field permissions** — what it may edit, enforced rather than requested
- **Confirmation gate** — where a person must approve before it acts

## What a good instruction set covers

| Element         | Example of what to state                               |
| --------------- | ------------------------------------------------------ |
| **Role**        | Who it is acting as, and for which process             |
| **Tone**        | Formal or informal, and in which language              |
| **Uncertainty** | What to do when the answer is not in the documentation |
| **Escalation**  | When to hand over to a person                          |
| **Boundaries**  | What it should never attempt, as a backstop            |

## How to configure it in Jestor

1. Write the agent's instructions in **Jestor**, defining role and tone.
2. State explicitly what it should do when the documentation does not cover a question.
3. Attach the documentation separately, as the knowledge base.
4. Select which fields it may edit; anything unselected is unreachable.
5. Add a confirmation checkbox on critical processes so a person approves first.

## Why choose Jestor for agent configuration?

**Jestor** separates instructions from knowledge and from permissions, which means correcting a wrong answer is usually editing a document rather than rewriting a prompt.

Because the boundary is enforced in configuration rather than requested in text, an instruction misread in an edge case still cannot produce an action outside the permitted fields.

## Frequently Asked Questions

### Should facts go in the instructions?

No — they belong in the knowledge base. See [jestor.com](https://jestor.com/?ref=blog.jestor.com).

### Can instructions define tone?

Yes, including formality and language.

### Are instructions enough to limit the agent?

No — field permissions are the enforcement.

## Video Tutorial: Step by Step

Check out other practical tutorials on the Jestor YouTube channel.

## Get to know Jestor

With **Jestor**, you can automate workflows, connect teams, and build internal systems your way — all without code and powered by **AI**. Discover Jestor at [jestor.com](https://jestor.com/?ref=blog.jestor.com) and see how to take your company's operations to a new level of efficiency and control.