Skip to main content
After generating a Sentinel (or opening an existing one), you can refine how it evaluates requests from the Sentinel edit page.

​
Core fields

  • Name: your label for what the Sentinel checks.
  • Severity: the Portal uses severity (INFO / WARN / ERROR / FATAL) to communicate how serious the issue is and to drive defaults in evaluation workflows (for example, higher-severity sentinels are preselected more often).

​
Evaluation directive

The Evaluation section is the main definition of what the Sentinel checks.
  • Directive: the natural-language instruction describing what the Sentinel should validate.
  • Examples: optional labeled examples you can add to anchor behavior.
    • Each example has a type like PASS, FAULT, or NA and a text body.

​
Qualification (optional)

Some sentinels include a Qualification section, which contains:
  • Type: STATIC or LLM
  • Directive: how to qualify the input before evaluation
If the section is empty, the Portal may offer a Generate / Regenerate workflow (see Regenerating Sentinels).

​
Correction (optional)

Some sentinels include a Correction section, which contains:
  • Type: STATIC or LLM
  • Directive: how to produce a suggested correction when a FAULT is detected

​
Include options

The Include Options toggles control what context is included when the Sentinel evaluates:
  • System Prompt
  • Context
  • Input
  • Tool Calls
  • JSON Schema

​
Save changes

When you’re satisfied, click Update Sentinel to persist the configuration.