> ## Documentation Index
> Fetch the complete documentation index at: https://docs.waterr.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Create a guardrail

> Register a guardrail on your account. Define once, then attach to any number of scenarios with [`Attach a guardrail to a scenario`](#tag/Scenario-Guardrails/operation/AttachGuardrailToScenario). Editing the definition later propagates to every attached scenario from its next meeting.



## OpenAPI

````yaml POST /guardrails
openapi: 3.0.0
info:
  title: WaterrAI API
  version: 1.0.0
  description: >-
    REST API for the WaterrAI meeting platform — create scenarios, spin up
    AI-driven meetings, fetch transcripts and analyses.
  contact:
    name: Support
    email: harshit@waterr.ai
servers:
  - url: https://api.waterr.ai/v1
    description: Production
security:
  - apiKeyAuth: []
tags:
  - name: Users
  - name: Personas
  - name: Voices
  - name: Scenarios
  - name: Goals
  - name: Meetings
  - name: Analyses
  - name: Recordings
  - name: Transcripts
  - name: Tools
    description: >-
      Account-scoped custom function (tool) definitions the LLM can call
      mid-meeting.
  - name: Scenario Tools
    description: >-
      Per-scenario attachments linking a scenario to one or more tool
      definitions.
  - name: Guardrails
    description: >-
      Account-scoped named prompt fragments appended to a scenario's system
      prompt.
  - name: Scenario Guardrails
    description: Per-scenario attachments linking a scenario to one or more guardrails.
  - name: Meeting Skills
    description: Account-scoped invokable instruction packs the agent loads mid-meeting.
  - name: Scenario Meeting Skills
    description: Per-scenario attachments linking a scenario to one or more skills.
  - name: Memories
    description: >-
      Facts your agents remember, in containers named by user_memory_tag and
      shared across agents.
paths:
  /guardrails:
    post:
      tags:
        - Guardrails
      summary: Create a guardrail
      description: >-
        Register a guardrail on your account. Define once, then attach to any
        number of scenarios with [`Attach a guardrail to a
        scenario`](#tag/Scenario-Guardrails/operation/AttachGuardrailToScenario).
        Editing the definition later propagates to every attached scenario from
        its next meeting.
      operationId: CreateGuardrail
      requestBody:
        required: true
        content:
          application/json:
            schema:
              type: object
              required:
                - guardrail_name
                - guardrail_prompt
              properties:
                guardrail_name:
                  type: string
                  pattern: ^[a-zA-Z0-9_-]{1,64}$
                  description: Unique per account.
                guardrail_prompt:
                  type: string
                  maxLength: 10000
                active:
                  type: boolean
                  default: true
            example:
              guardrail_name: no_competitors
              guardrail_prompt: >-
                Only mention products within Our Company Inc. during
                conversations; never discuss competitors.
      responses:
        '201':
          description: Guardrail created.
          content:
            application/json:
              schema:
                $ref: '#/components/schemas/Guardrail'
        '400':
          description: >-
            Invalid payload — see the `errors` array for the specific field
            failures.
        '401':
          description: Unauthorized — missing or invalid API key.
        '409':
          description: A guardrail with that name already exists on this account.
      security:
        - apiKeyAuth: []
components:
  schemas:
    Guardrail:
      type: object
      description: >-
        A named prompt fragment. Every guardrail attached to a scenario is
        appended to that scenario's system prompt as a bullet under a `##
        Guardrails` heading, so the persona treats it as a non-negotiable rule
        for the whole conversation.
      required:
        - id
        - guardrail_name
        - guardrail_prompt
        - active
      properties:
        id:
          type: string
          format: uuid
          example: 9b1f3d70-2a4e-4c8b-8f1a-6d5c4b3a2e10
        membership_id:
          type: string
          format: uuid
          example: a1b2c3d4-e5f6-7890-abcd-ef0123456789
        guardrail_name:
          type: string
          maxLength: 64
          pattern: ^[a-zA-Z0-9_-]{1,64}$
          description: >-
            Handle for the rule. Unique per account. Used in the API and in your
            logs — it is never shown to the persona or the participant.
          example: no_competitors
        guardrail_prompt:
          type: string
          maxLength: 10000
          description: >-
            The rule itself, appended verbatim to the system prompt. Write it as
            a direct instruction. Encrypted at rest.
          example: >-
            Only mention products within Our Company Inc. during conversations;
            never discuss competitors.
        active:
          type: boolean
          default: true
          description: >-
            Account-level kill-switch. `false` removes the rule from every
            scenario at once.
          example: true
        scenarios:
          type: array
          description: >-
            Scenarios this guardrail is attached to. Returned only on the
            single-guardrail read.
          items:
            type: object
            properties:
              id:
                type: string
                format: uuid
              name:
                type: string
              active:
                type: boolean
              position:
                type: integer
        created_at:
          type: string
          format: date-time
          example: '2026-09-09T12:34:56.000Z'
        updated_at:
          type: string
          format: date-time
          example: '2026-09-09T12:34:56.000Z'
  securitySchemes:
    apiKeyAuth:
      type: http
      scheme: bearer
      bearerFormat: wai_...
      description: >-
        API key from waterr.ai/settings?tab=api-keys. Send as `Authorization:
        Bearer wai_<your_key>`.

````

This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.