Workflows · Tools
Guardrails
Checks incoming content and earlier outputs before the workflow continues.
What it does
Looks for content that matches the rules or optional AI policy you define, then blocks, warns, or redirects the workflow.
When to use it
Use it to prevent unsafe replies, protect sensitive information, enforce tone, or route exceptional cases to a safer path.
How to use it
- 1Add Contains, Does not contain, or Regular expression rules.
- 2Choose whether each rule checks the incoming message, earlier outputs, or both.
- 3Choose Block, Warn, or Redirect.
- 4If you choose Redirect, connect the redirect path.
Settings
- Structured rules.
- Optional AI policy.
- Checks: Incoming message, Earlier outputs, or All.
- Action: Block, Warn, or Redirect.
Tips
- • Use deterministic rules for requirements that must always be enforced.
- • Make the redirect path safe and easy for a person to review.