Skip to content

Sentence Count Guardrail

The Sentence Count Guardrail validates the number of sentences in a request or response. Use it to enforce policies on the structural complexity of prompts or responses — for example, requiring responses to be at least a certain length, or capping how many sentences a user prompt can contain.

Configuration

All parameters are under Advanced Settings. Configure a request section, a response section, or both.

Request / Response Parameters

Each section (request and response) has the following parameters:

Parameter Required Default Description
min Yes 0 Minimum number of sentences required.
max Yes 0 Maximum number of sentences allowed. Set to 0 to disable the upper limit.
jsonPath No — JSONPath expression to extract a specific field for validation (e.g., $.messages[0].content). If not set, the entire payload is evaluated.
invert No false When true, the guardrail intervenes when the sentence count is within the configured range instead of outside it.
showAssessment No false When true, the intervention response includes details about the violation.

Add This Guardrail

  1. Navigate to AI Workspace > LLM Providers or LLM Proxies.
  2. Click on the provider or proxy name.
  3. Go to the Guardrails tab.
  4. Click + Add Guardrail and select Sentence Count Guardrail from the sidebar.
  5. Expand Advanced Settings and configure the request and/or response sections.
  6. Click Add (for providers) or Submit (for proxies).
  7. Deploy the provider or proxy to apply the changes.

Example: Limit Prompt to 3 Sentences

The following configuration blocks requests where the user message contains more than 3 sentences.

Advanced Settings — request:

Parameter Value
min 1
max 3
jsonPath $.messages[0].content
invert false
showAssessment true

Intervention response:

{
  "type": "SENTENCE_COUNT_GUARDRAIL",
  "message": {
    "action": "GUARDRAIL_INTERVENED",
    "interveningGuardrail": "Sentence Count Guardrail",
    "direction": "REQUEST",
    "actionReason": "Violation of applied sentence count constraints detected.",
    "assessments": "Violation of sentence count detected. Expected between 1 and 3 sentences."
  }
}