Write your agent's instructions
An agent's instructions live in AGENTS.md. Agenta sends this file to the agent at the start of every session. This guide shows you how to edit it, write a first version, and add corrections. To learn what belongs in instructions rather than a file or skill, see Instructions.
Open the editor
- Go to
Agentsin the sidebar and pick your agent. - In the Configuration column, click the
AGENTS.mdrow under Instructions.
Write a first version
A first version for a marketing coworker:
# Role
You are the marketing coworker at Northwind, which sells scheduling software to
independent dental practices. You draft copy, review drafts I write, and answer
questions about what we have shipped.
## Audience
Practice owners and office managers. Busy, not technical. They buy to save the front
desk time.
## How to talk to me
Answer in under 150 words unless I ask for a draft. No preamble and no restating my
question.
## Hard limits
Never publish or send anything. Show me the draft and wait.
Then click Save. The dialog closes and the configuration saves itself as a new version: the state dot in the agent header reads Saving…, then Saved.
Add a correction
Improve the instructions as you work with the agent. When it gets something wrong, add the rule that would have prevented the mistake.
For example, the agent drafts a LinkedIn post that opens with "In today's fast-paced world" and includes three emoji. Reopen AGENTS.md and add the correction to the existing tone section:
## How to talk to me
Answer in under 150 words unless I ask for a draft. No preamble and no restating my
question. In social copy: no emoji, and never open with a scene-setting sentence.
Open with the claim.
Save again. Each save creates a version, and the vN chip in the agent header carries the summary of what changed.
Ask the agent to edit the instructions
You can also describe the change in chat. When Playground build kit is on in the Advanced dialog, the agent can update its own configuration, including its instructions. This setting is on by default.
Next
- Control what an agent can do covers the limits that are settings rather than sentences.