Prompt Library & Templates

Reusable, parameterized prompts with rich input forms, variables, conditional logic, and AI-generated options.

Findable includes a full-featured prompt template system for creating reusable, parameterized prompts with rich input forms.

Prompt Templates

A prompt template (IPromptTemplate) consists of:

• Content — Markdown text with {{variableName}} placeholders
• Variable Specs — Rich input definitions (IPromptVariable[]) that describe how each placeholder should be rendered as a form control
• Tags — For organization and search
• Scope — Personal (creator-only) or Shared (ACL-controlled)
• Groups — Templates can be organized into prompt groups with drag-and-drop ordering

Variable Types

Each variable has a type that determines the input control rendered in the form:

Basic Types (always available):

Type	Control	Description
text	TextField	Single-line text input
multiline	TextArea	Multi-row text with configurable rows
number	Numeric input	With optional min/max constraints
select	Dropdown	Single-select from options list
multiselect	Multi-select chips	Multiple selections from options list
boolean	Toggle switch	True/false input
date	Date picker	Calendar date selection
file	File upload	Single or multiple file attachments
principal	Azure AD picker	User and/or group selection

Advanced Types (require advanced mode toggle):

Type	Control	Description
range	Slider	Numeric range with step control
hierarchy	Tree builder	Nested structure input (org charts, nested lists)
datetime	Date+Time picker	Combined date and time selection
time	Time picker	Time-only selection
hidden	None (invisible)	Pass data without UI — injected silently
password	Masked input	Sensitive text entry
table	Editable table	Structured or unstructured tabular data entry
repeater	Dynamic list	Add/remove rows of sub-field groups

Select Option Formats

Select and multiselect variables support three option formats:

• Simple strings — ["Option A", "Option B"]
• Grouped options — { group: "Fruits", items: ["Apple", "Banana"] } for hierarchical dropdowns
• Rich options — { value: "opt1", label: "Option 1", description: "Help text", icon: "CheckCircle", disabled: false } for metadata-enhanced options

Dynamic Options Sources

Instead of static options, select/multiselect variables can populate from system objects at runtime:

Source	Populates From
models	AI Model Endpoints
searchEndpoints	Azure AI Search Endpoints
dataPlatformConnections	SQL/Database Connections
mcpServers	MCP Server Registry
tools	Tool Providers

AI-Generated Options

Individual select/multiselect variables can have AI generate their options at runtime:

• aiPopulate — Enable AI-generated options for this variable
• aiPrompt — Prompt sent to AI (supports {{variable}} references from flow context)
• aiModelEndpointId — Optional model override
• aiOptionsCount — Number of options to generate (default: 5)

Conditional Logic

Variables support dynamic show/hide/require/disable rules based on other field values:

{
  "conditionals": [{
    "field": "department",
    "operator": "equals",
    "value": "Engineering",
    "action": "show"
  }]
}

Supported operators: equals, not_equals, contains, not_contains, greater_than, less_than, is_empty, is_not_empty, in, not_in

Supported actions: show, hide, require, disable, enable

Layout & Visual Options

• Width — full, half, third, quarter for responsive grid layout
• Variant — default, radio, chips, buttons for select/multiselect/boolean display
• Ordering — Explicit order field for display sequence
• Grouping — Visual grouping via the group field

Text Validation

Text and multiline variables support validation presets: email, url, phone, ipv4, ipv6, subnet, uuid, alphanumeric, numeric, or custom (regex pattern).

Chat-Specific Prompts

Each chat can be assigned a prompt group from the Prompt Library via the promptGroupId field on the chat configuration. This surfaces the group’s prompt templates in the chat’s prompt selector, giving each chat a curated set of prompts without duplicating data.