Project

Name the project, theme the learner experience and wire the services. Everything here lands in the exported config.

Identity

Project name (internal, used for the save file)

Client name (shown to learners)

Product name (the title learners see)

Counterpart noun (guest, patient, caller, customer)

Session noun (what a practice is called)

Logo text (2 to 3 letters, or upload below)

Logo image (small PNG or SVG, optional)

Theme

Accent

Accent dark

Ink (text)

Paper (page)

Surface (cards)

Warn

Report style

Report title

Scenario Sim

The learner sees this palette across the whole experience.
Start a practice

Session settings

Pass mark %

Max attempts (0 = unlimited)

Report score on

Voice mode Text mode Spoken debrief on the report

Manager contact line (shown on the report, optional)

Services

ElevenLabs agent ID

ElevenLabs API key

Anthropic API key (pilot: ships in the zip)

Scoring proxy URL (production: overrides the key)

Coach voice ID (optional, for the spoken debrief)

xAPI (optional)

LRS endpoint

Username / key

Password / secret

Leave xAPI blank to skip it. SCORM reporting to the LMS always happens.

Knowledge

Drop in the client's world as plain text or documents. Structure it with AI, then tidy the result in the editors. This becomes the counterpart's grounding, the in-session support panels and the assessor's source of truth.

Organisation

Organisation name

Source material

Drop Word docs, PDFs or text files here, or browse

Or paste anything (price lists, policies, procedures, SOP extracts)

Reads every source and proposes reference sections, channels, policies and exclusions. Nothing applies until you approve it.

Reference sections

Price lists, procedures, product info. Lists are one item per line. Tables use pipes: first line is the headers, each line after is a row.

Channels

Where conversations happen: front desk, phone, drive-thru, ward round. Each has staged steps the assessor can mark. One stage per line as Label :: what good looks like.

Policies

House rules, marked directly. One per line.

Exclusions

Things we do not do. The counterpart may push for them. The assessor never penalises a polite decline that offers the listed alternative.

Difficulty

Voices

Named voice slots scenarios can point at (for example female-older). Paste ElevenLabs voice IDs, or fetch your voice library.

Rubrics

One rubric pack per scenario type. Weight what matters 1 to 3, and keep the done-well descriptions concrete: they steer the assessor.

Pack label

Pack key (scenarios reference this)

Criteria

ID	Criterion	Weight	Applies to	What done well looks like

Upload CSV

Auto-fail conditions

One per line. Any of these caps the score below the pass mark regardless of everything else.

Scenarios

The library learners pick from. Build with AI from a rough brief, then fine-tune every field.

Test run

Launches the real learner build in a new tab, exactly as it will export, using your keys. Voice needs the ElevenLabs agent wired and overrides enabled in its Security tab.

Scenario

Preselect difficulty

Mode

Opens in a new tab so the microphone works. Allow pop-ups for this page. SCORM calls no-op outside an LMS.

Export

Generates the learner-only SCORM 1.2 package. Admin does not exist in the exported code, not even hidden.

Pre-flight checks

Package

Zip file name

SCORM course title (shows in the LMS)

LMS reminders: launch in a new window (the microphone is blocked in iframes), allow-list elevenlabs.io, cdn.jsdelivr.net, api.anthropic.com (or your proxy) and fonts.googleapis.com, and if voice sounds wrong check all four override switches in the agent's Security tab.