From Tools to Teacher Built Teammates No Code Pedagogical Plugin Authoring with LearnAdapt Agentic Studio and PedOS 1.1 Lumina

Jul 16, 2026

Upload by Skyler

Teachers and researchers need to adapt educational AI to local goals, yet customization and study usually require coding expertise. LearnAdapt Agentic Studio on PedOS 1.1 Lumina provides a no-code authoring and governed runtime for educational AI plugins. A non-coder specifies a learning interaction in plain English; the system generates a previewable plugin artifact, performs safety checks, and supports admin review. PedOS deploys approved plugins, while telemetry is tightly gated to authenticated users running approved, installed plugins, enabling controlled evidence capture.

LanguageEnglish

CategoryTechnology

FormatPDF

Pages3

Reading Time8 min

Motivation

System Presentation

Agentic Studio

Security and Review

PedOS 1.1 Lumina

Demonstrated Functionality

Originality, Maturity, and Future Work

arXiv :2607 .09674v 1 [ cs .CY] 5 Jun 2026 From Tools to Teacher-Built Teammates: No-Code Pedagogical Plugin Authoring with LearnAdapt Agentic Studio and PedOS 1.1 Lumina Nizam Kadir Singapore University of Technology and Design [nizam_kadir@mymail.sutd.edu.sg](nizam_kadir@mymail.sutd.edu.sg) Abstract. Teachers and researchers need to adapt educational AI to local goals, but most systems remain difficult to customize or study without coding expertise. We present LearnAdapt Agentic Studio on PedOS 1.1 Lumina, a no-code authoring and governed runtime environment for educational AI plugins. A non-coder describes a desired learning interaction in plain English; the system prepares a previewable plugin artifact, runs safety checks, and supports submission for review. PedOS then deploys approved plugins into a directory for installation. Crucially, telemetry is strictly gated to authenticated users running approved plugins. The demo shows the complete lifecycle from prompt to governed evidence capture, shifting from fixed tools to teacher-built teammates. Keywords: Educational AI · no-code authoring · agentic systems · plugin ecosystem Demo Link [https://www.youtube.com/watch?v=PjIRnAKUxWc](https://www.youtube.com/watch?v=PjIRnAKUxWc) 1 Motivation The AIED 2026 theme raises a practical question: who shapes the AI teammate? [1]. In many deployments, educators are unable to author learning interactions [2] or control evidence capture [3], remaining users of fixed systems [4] . LearnAdapt addresses this by treating AI functionality as governable plugins. A non-coder describes an activity in plain English, previews the plugin, and submits it for review. After approval, it is distributed via PedOS 1.1 Lumina. This design ensures pedagogical utility, enforces safety reviews, and captures research evidence through explicit, controlled telemetry pathways. 2 N. Kadir 2 System Presentation LearnAdapt Agentic Studio and PedOS 1.1 Lumina form a no-code authoring and governed runtime pipeline. Agentic Studio is the creator interface where an educator’s prompt acts asa client brief for an agent development team. The workflow begins with deterministic preflight checks, followed by specialized agents handling pedagogy, UX, and architecture in parallel. Next, build and evidence design occur simultaneously, culminating in QA, preview iteration, and release packaging. This multi-agent pipeline generates runtime PHP/JS/CSS artifacts that execute securely via platform routing. Direct access is blocked, and creators preview interactions before admin review. Fig. 1. The LearnAdapt Agentic Studio authoring interface. Educators describe the desired learning interaction in plain English, and the agentic pipeline generates a deployable, governable plugin through transparent build stages. Security and review forms the governance layer. Validation includes PHP syntax, API boundaries, and static application security testing (SAST) . Admin Teacher-Built Teammates with Agentic Studio and PedOS 3 approval re-runs these checks before deployment, using SAST scores alongside policy status. PedOS 1.1 Lumina is the runtime layer. Approved plugins are versioned and surfaced in the Plugin Directory for users to preview, install, and access. Strict runtime safeguards isolate plugin execution to prevent platform-wide failures. Telemetry gating links interaction to evidence. Preview modes use mock telemetry. Events are persisted only for authenticated users running active, installed plugins via approved endpoints. 3 Demonstrated Functionality The demo presents a full plugin lifecycle. An educator prompts Agentic Studio for a science retrieval-practice plugin. After previewing the generated artifact, they submit it. An admin reviewer inspects the artifact, preview, SAST score, and policy status, then approves it. The Plugin Directory displays the deployed package, allowing learners to install and use it. The installed plugin e

FAQs

How does a non-coder create an educational AI plugin in LearnAdapt Agentic Studio?

The educator describes a desired learning interaction in plain English. Agentic Studio then generates a previewable plugin artifact, runs preflight checks, and supports submission for admin review.

What governance and security steps are applied before deployment?

Governance includes validation such as PHP syntax checks, API boundary checks, and static application security testing (SAST). Admin approval re-runs these checks before deployment.

How is telemetry for evidence capture controlled?

Telemetry is strictly gated to authenticated users who run active, installed plugins through approved endpoints. Preview modes use mock telemetry, and persisted events occur only for approved plugin execution.

Jul 16, 2026

Upload by Skyler