Core Features

Learner Integration

SlideMind connects your e-learning modules to AI through two integration methods. Which one is used depends on the slide’s widget mode.

Three modes

Every slide is configured with one of three modes:

Mode	Behavior	Integration method	Use case
Q&A (default)	Single-turn: the learner sends one message and gets one response.	JavaScript code	Quick feedback, quiz evaluation, one-off explanations
Chat	Multi-turn conversation. Messages accumulate in a scrollable history.	HTML Widget	Tutoring sessions, guided exploration, Socratic dialogue
Capture	No AI involved. Saves e-learning variables to learner memory silently.	JavaScript code	Collecting learner data for use in later slides

Note: Capture mode requires a Pro or Agency plan.

JavaScript integration (Q&A and Capture)

In Q&A and Capture modes, SlideMind generates JavaScript code that you paste directly into your e-learning editor (Storyline trigger, Captivate action, etc.). The learner interacts with the editor’s own interface — SlideMind runs in the background.

How it works:

1. The learner types in a text entry field in your e-learning module (e.g. a Storyline text entry bound to a userMessage variable).
2. A trigger runs the SlideMind JavaScript code, which reads the variable, sends it to the SlideMind REST API on your WordPress server, and writes the AI response back to an editor variable (e.g. aiResponse).
3. The e-learning module displays the response using its own text box.

The learner never sees an iframe, a widget, or anything outside the e-learning module. You have full control over the visual design.

For details on each authoring tool’s syntax and setup steps, see the Authoring Tool Guides.

HTML Widget (Chat mode)

For multi-turn conversations, SlideMind provides an HTML Widget — an HTML page hosted on your WordPress site, embedded as an iframe or web object in your e-learning module. The widget has its own conversation UI because e-learning editors can’t manage a scrollable message history natively.

The HTML Widget is required for Chat mode, and optionally available for Q&A mode if you prefer its built-in UI over designing your own in the editor.

Widget interface

• Header — Displays the slide title (or a custom title). In Chat mode, a reset button lets the learner start over. The header can be hidden entirely via appearance settings.
• Message area — User messages appear on the right (in the primary color), AI responses on the left. Errors are centered in red.
• Typing indicator — An animated three-dot indicator appears while waiting for the AI response.
• Input field — Auto-resizing textarea. Press Enter to send, Shift+Enter for a new line.
• Themes — Light (default) and dark themes. Fully styled with CSS variables (--sm-*). Supports prefers-reduced-motion.
• Appearance customization — Colors (accent, AI bubble, AI text), border radius, header visibility, and title are all customizable per slide or globally in Settings. A Customize widget button in the Integration Code card opens a modal with live preview.

Chat mode options

When a slide uses Chat mode, you can configure:

Option	Description	Default
Turn limit	Maximum number of exchanges before the conversation closes. Set to 0 for unlimited.	5
Closing message	Text displayed when the turn limit is reached (e.g. “Thank you for this exchange!”).	—
Graceful close	When enabled, an instruction is injected into the system prompt on the last turn, asking the AI to naturally wrap up the conversation. A subtle banner (instead of an AI bubble) marks the end.	Off

A cost projection below these options estimates the total cost of a full conversation based on the selected model’s pricing and the turn limit.

Consent banner (optional)

Enabled via SlideMind > Settings > Advanced > Widget consent. When active, an overlay appears before the learner can interact:

• An information message explaining that messages will be sent to an AI provider
• A “I understand, continue” button
• Once accepted, the conversation becomes accessible

Useful for GDPR/Loi 25 compliance. You can disable it if your site already manages consent globally.

URL parameters

The widget URL accepts several parameters for customization:

Parameter	Description	Default	Example
sm_widget	Activates widget rendering (required)	—	1
config	Slide configuration ID (required)	—	abc123
theme	Visual theme	light	dark
color	Accent color — header, buttons, user bubbles (URL-encoded hex)	#4f46e5	%234f46e5
ai_color	AI bubble background color (URL-encoded hex)	#f1f5f9 (light) / #1e293b (dark)	%23e2e8f0
ai_text_color	AI bubble text color (URL-encoded hex)	#1e293b (light) / #e2e8f0 (dark)	%23334155
title	Header title	Slide title	My+Chat
header	Show or hide the header	1	0
radius	Border radius in pixels for bubbles and input	12	8
sm_user	Learner WordPress user ID (for memory features)	0	42

All appearance parameters follow a cascade: URL parameter > slide-level override > global Settings default > hardcoded fallback. You can set defaults in SlideMind > Settings > Defaults, override per slide via the Customize widget button, and further override via URL parameters.

Example URL:

https://yoursite.com/?sm_widget=1&config=abc123&theme=dark&color=%23e11d48&title=Roleplay&header=0

---

Sandbox

The Sandbox is a testing environment where you can try your configurations live, see exactly what the AI receives, and debug issues — without affecting your published slides.

Go to SlideMind > Sandbox to access it.

Two modes

Configuration mode

Load an existing slide configuration and test it:

1. Select a slide from the dropdown.
2. The system prompt, model, temperature, max tokens, and variables are auto-filled from the saved configuration.
3. Modify any value for testing — changes stay in the Sandbox and are not saved to the database.
4. Type a test message and send it.

A collapsible prompt reference shows the full system prompt for quick review.

Conversation mode (manual)

Build a configuration from scratch:

1. Write a system prompt directly.
2. Select a model from the dropdown.
3. Add variables manually in the dynamic variable table (add/remove rows).
4. Variables are auto-detected from {{placeholders}} in the prompt.

In this mode, the conversation history persists (stored as a WordPress transient for 24 hours, per user). You can test multi-turn interactions this way.

Debug panel

After each request, the debug panel shows:

Field	Description
Original prompt	The raw system prompt with {{variable}} placeholders
Rendered prompt	The final prompt with all variables injected
Variables found	Variables detected in the prompt
Variables used	Variables that were successfully resolved
Variables missing	Variables in the prompt that had no value
Tokens	Prompt tokens + completion tokens + total
Estimated cost	Based on the model’s pricing
Response time	In milliseconds

Click any {{variable}} tag in the debug panel to see its resolved value in a modal.

Additional features

• AI prompt generation — The “Generate with AI” button works here too, so you can draft prompts without leaving the Sandbox.
• Export configuration — Download the current Sandbox state as a JSON file.
• Session management — The conversation persists for 24 hours. Sessions are cleaned up on logout.

Multi-model comparison (Pro)

Send the same prompt to 2-3 models simultaneously and compare them side by side.

Aspect	Details
Models	Up to 3 models in parallel
Metrics	Response quality, tokens used, cost, response time
Export	Results exportable as JSON

Compare cost, speed, and response quality in a single test to pick the right model for your use case.

---

Templates

Templates are reusable prompt configurations that save time when creating slides. Instead of writing a system prompt from scratch every time, pick a template and customize it.

Go to SlideMind > Templates to manage them.

Default templates

SlideMind ships with 6 built-in templates, organized by widget mode:

Q&A mode (single-turn):

Template	Slug	Description
Quiz Feedback	quiz	Personalized, constructive feedback on quiz answers
Open Question Feedback	open-question	Short, personalized feedback on a free-text response
Formative Evaluation	formative-eval	Evaluates knowledge and returns a score, feedback, and verdict

Chat mode (multi-turn):

Template	Slug	Default turns	Description
Socratic Tutor	socratic	8	Guides the learner through open-ended questions rather than giving answers directly
Roleplay	roleplay	10	Interactive scenario simulations and role-based practice
Guided Exploration	guided-exploration	12	Guides the learner through a topic via a structured conversation

Each template is tied to a widget mode. In the Wizard, only templates matching the selected mode (Q&A or Chat) are shown.

Default templates cannot be deleted, but they can be edited. Use the Reset button to restore a default template to its original state.

Creating a template

Click Add Template and fill in:

Field	Description
Title	A descriptive name
Slug	URL-friendly identifier (auto-generated from the title if left blank)
Icon	An emoji (selected via emoji picker)
Description	What this template does — shown as a tooltip in the Wizard
System prompt	The reusable prompt, with {{variable}} placeholders
Variables	Auto-detected from the prompt
Temperature	Suggested temperature for this template
Max tokens	Suggested max tokens

Favorites

Mark up to 6 templates as favorites by clicking the star icon. All 6 default templates are favorited out of the box. Favorites get special treatment in the Wizard:

• Favorites are displayed as clickable cards with their emoji, title, and description.
• Non-favorites appear in a dropdown labeled “Other templates”.
• If no favorites are set, all templates are shown as cards (backward-compatible).
• Only templates matching the current widget mode (Q&A or Chat) are displayed.

Manage favorites in the template list or in SlideMind > Settings.

Import / Export

Share templates between sites or with other users:

• Export — Downloads the template as a JSON file (format version sm_template_version: 1.1).
• Import — Upload a JSON file. The schema is validated before creating the template.

Template JSON format:

{
  "sm_template_version": "1.1",
  "name": "My Template",
  "slug": "my-template",
  "description": "What this template does",
  "icon": "🎓",
  "interaction_type": "socratic",
  "system_prompt": "You are a tutor helping {{learner_name}} learn about {{topic}}...",
  "temperature": 0.7,
  "max_tokens": 1024,
  "variables": ["learner_name", "topic"]
}

---

Code Generator

After creating a slide configuration, the code generator produces ready-to-paste integration code for your e-learning editor.

Where to find it

The code generator appears in three places:

1. Slide list page — Click the copy button on any slide row to copy the integration code directly.
2. Slide edit page — In the “Integration code” card on the right sidebar.
3. Wizard step 4 — After creating a configuration, the integration code is shown in the review.

How it works

1. Select your authoring tool — Storyline, Captivate, Lectora, DominKnow, iSpring, or HTML Widget (iframe).
2. Copy the code — Click the copy button. The code is ready to paste.
3. Paste in your authoring tool — Follow the tool-specific instructions (see the Authoring Tool Guides).

How the generated code works

The generated code is a short bootstrap snippet (~12 lines) that dynamically loads the SlideMind runtime (sm-slide.js) from your WordPress server. The runtime is an external JavaScript file that handles learner identification, memory management, API calls, and editor variable read/write.

var s = document.createElement('script');
s.src = 'https://yoursite.com/.../sm-slide.js?v=0.5.1';
s.onload = function() {
    SlideMind.run({
        configId:  "abc123",
        baseUrl:   "https://yoursite.com/wp-json/slidemind/v2/",
        mode:      "qa",
        editor:    "storyline",
        inputVar:  "userMessage",
        outputVar: "aiResponse"
    });
};
document.head.appendChild(s);

The sm-slide.js runtime handles:

• Learner identification — Automatic cascade: sm_user URL param > SCORM 2004 > SCORM 1.2 > xAPI actor > anonymous session (UUID in localStorage).
• Memory — SM_Memory.save() and SM_Memory.load() singleton, initialized automatically.
• Chat request — XHR POST to /wp-json/slidemind/v2/chat with message, variables, and learner identity. Handles both plain text and structured JSON responses.
• Editor adapters — Runtime variable read/write for each authoring tool (Storyline, Captivate, Lectora, DominKnow, iSpring).
• Captivate polling — Automatic retry until cpAPIInterface is available (up to 10 seconds).

Mode-specific behavior

Mode	Generated code
Q&A	Bootstrap snippet that loads sm-slide.js with mode: "qa". Reads the input variable, sends to API, writes the response variable (and any return variables).
Chat	Always generates an HTML Widget embed (iframe), regardless of the selected editor. Multi-turn conversations require the widget’s built-in UI.
Capture	Bootstrap snippet with mode: "capture". Saves the listed memoryVars to learner memory. No AI request. The HTML Widget option is disabled because the iframe can’t access the editor’s variables (see why).

Customization options

The code generator lets you customize:

• Editor — Which e-learning tool you’re using (determines the runtime adapter used by sm-slide.js)
• Variable mapping — The names of the editor variables for input and response. SlideMind auto-generates unique names per slide (e.g. userMsg_s3_quizFeedback, aiResp_s3_quizFeedback) to avoid conflicts when a project contains multiple slides. You can rename them in the slide edit page; a warning appears if a name is already used by another slide in the same course.
• Widget appearance — Click the Customize widget button (below the copy button) to open the appearance modal with live preview. Settings are saved per slide and injected into the generated widget URL. See URL parameters for the full list of customizable properties.

For details on each authoring tool’s syntax and setup steps, see the Authoring Tool Guides.

---

Wizard

The Wizard is a guided 4-step process for creating a new slide configuration. Non-technical users can follow it step by step; experienced users can move through it quickly.

Go to SlideMind > Add to launch the Wizard.

The Wizard is covered in detail in the Getting Started guide. Below is a summary of each step.

Step 1 — Hierarchy & Mode

Set up where the slide lives (course, module) and how it behaves (Q&A, Chat, or Capture).

• Select an existing course and module from the dropdowns, or leave them empty to create an independent slide.
• The mode selector uses radio cards with icons and descriptions for clarity.

Step 2 — AI Model

Choose the LLM model, temperature, and max tokens.

• Skipped automatically in Capture mode.
• Models are grouped by provider (Anthropic, OpenAI, Google, Meta, etc.) with estimated cost per interaction.
• Pre-filled from your defaults in SlideMind > Settings.

Step 3 — System Prompt / Variables

Write or generate the system prompt.

• Template selector — Pick a template to start from. Only templates matching the selected mode (Q&A or Chat) are shown. Favorites appear as cards, others in a dropdown.
• Variable bar — Clickable variable tags (system variables in blue, custom in green). Click to insert.
• @mention autocompletion — Type @ in the prompt editor for variable suggestions (powered by Tribute.js).
• Generate with AI — Opens a modal where you describe your pedagogical objectives, and an AI drafts the system prompt for you.

Response example — An optional field where you provide a concrete example of a good AI response. This helps the AI match the desired format, tone, and detail level. See Response Example.

Mode-specific sections:

• Chat — Turn limit, closing message, graceful close toggle, cumulative cost projection.
• Q&A — Return variables (name + type: text, number, boolean, enum).
• Capture — Variable definitions (memory key, scope, editor variable name) instead of the prompt editor.

Advanced options (collapsible, all modes):

• Auto-save exchange (Q&A/Chat) — Save the learner’s input and/or the AI response to memory. See Auto-save exchange.
• Capture variables (Q&A/Chat) — Capture e-learning editor variables to memory alongside the AI interaction.
• Internal notes — Admin-only notes, never sent to the AI.

Step 4 — Review & Test

Review the complete configuration and test it live before saving.

• Summary — Full configuration at a glance.
• Prompt preview — Rendered prompt with test variable values (Q&A/Chat).
• Capture summary — Variables to be captured (Capture).
• Test — Send a test message to the AI with debug info (Q&A/Chat).
• Create — Save the configuration and go to the slide edit page.

---

AI Prompt Generation

Let an AI draft your system prompt based on your pedagogical objectives. This is useful when you’re not sure how to phrase instructions for the AI.

Where to access it

• Wizard step 3 — “Generate with AI” button below the prompt editor
• Slide edit page — Same button in the prompt section
• Sandbox — Available in conversation mode

How to use it

Click Generate with AI to open the generation modal. Fill in the fields:

Field	Max length	Example
Interaction type	100 chars	”Socratic tutor”
Subject	200 chars	”Non-violent communication”
Learning objective	300 chars	”Identify the 4 steps of NVC”
Tone	100 chars	”Warm and encouraging”
Learner level	50 chars	”Beginner”
AI model	—	Dropdown (defaults to your settings)
Available variables	Max 50	Multi-select from detected variables

Click Generate. The AI produces a complete system prompt using the selected variables, and inserts it into the prompt editor. You can then edit it freely.

Rate limiting

Prompt generation is limited to 10 requests per minute per user.

---

Internal Notes

Every slide has an Internal notes field (in the Advanced options section of the Wizard, or directly on the slide edit page). Notes are visible only to administrators — they are never sent to the AI or shown to learners.

Use them for reminders, context about why a slide was configured a certain way, or coordination notes between team members.

---

Next steps

• Getting Started — Installation and first slide setup
• Authoring Tool Guides — Integration code for each e-learning authoring tool
• Learner Memory — Persist data across slides, modules, and courses
• Variables — Variable system, profile variables, return variables
• Import / Export — Move configurations between WordPress sites
• Administration — Dashboard, analytics, settings, CORS