What are secrets in automation?
A secret is a credential a workflow needs, such as an API key or a password. It belongs in the tool's connection vault or a password manager, not in a prompt, a screenshot, or an SOP.
In plain English
If the text lets someone into an account, it is a secret. The automation tool usually has a lock icon for it. People still paste the key into a chat to ask why a call failed. That paste is the leak. Environment variables are a developer's version of the same drawer. You do not need the phrase. You need the drawer. A secret is anything that lets a stranger act as you: a key, a token, a password, a webhook address that triggers a job. Keep secrets in the tool's secret store. They do not go in prompts, screenshots, or the SOP. When you share your screen to film a how-to, cover them. Rotate one on purpose so you know the steps before you need them in a panic. Do not invent 'temporary' secrets in a personal password manager that nobody else can open when you are on a job site.
How an owner would use it
Put each secret in the connection screen that uses it. When someone who could see it leaves, replace it. Never ask a model to 'use this key.'
Common mistakes
A shared doc of passwords labeled 'do not share' will be pasted into a model. So will a key in a Zap's note field that the whole team can read. Rotating the key after a leak is required, and it does not erase a chat log that already stored it. Check retention after you rotate.
A worked example
A solar installer
The monitoring login stays in the password manager
This is an illustration, not a customer result. A tech wants the assistant to pull yesterday's production. The owner does not paste the portal password into the chat. They export the one day's numbers themselves and let the assistant format the customer note from those numbers.
Related terms
Prompts and guides
Questions owners ask
What is a secret in an automation?
A secret is a key or password the workflow uses. It should be stored in a vault, not in the prompt or the procedure doc.
How do I keep secrets out of AI chats?
Paste outputs, not credentials. If a call fails, redact the key from the error before you ask a model what it means.
What is an environment variable?
It is a named slot outside the code where a secret can sit for software to read. Same rule: the value does not belong in a chat.
Product guides are on AI apps for business.