OpenMAIC Releases v1.4.2 with Course-Level Undo and PBL v3 for Project-Based Learning

OpenMAIC (Open Multi-Agent Interactive Classroom), an open-source AI platform that transforms any topic or document into an immersive, multi-agent classroom experience, has released version 1.4.2 as of November 9, 2026. The project, hosted on GitHub by THU-MAIC, enables AI teachers and AI classmates to lecture, discuss, draw on a whiteboard, and engage learners in real time through generated slides, quizzes, interactive simulations, and project-based learning activities.

What OpenMAIC Does

• One-click lesson generation: Describe a topic or attach materials; the AI builds a full lesson in minutes.
• Multi-agent classroom: AI teachers and peers lecture, discuss, and interact with learners in real time.
• Rich scene types: Slides, quizzes, interactive HTML simulations, and project-based learning (PBL).
• Whiteboard & TTS: Agents draw diagrams, write formulas, and explain out loud.
• Export anywhere: .pptx preserving every scene, one offline .html file that plays the whole lesson, or a classroom ZIP for migration.
• Agent workbench integration: The OpenMAIC Skill works with OpenClaw, Codex, DeepSeek, WorkBuddy, and more — generate classrooms from Feishu, Slack, Telegram, 20+ messaging apps, or your IDE.

Recent Major Releases

v1.4.2 (2026-11-09)

• Course-level undo/redo: one undo/redo line per course covers manual edits, page changes, narration, and agent runs, with buttons in the workbench header and Cmd/Ctrl+Z.
• Director no longer retries a failing teacher indefinitely.
• Pro edits no longer leave an interactive page empty.
• `ALLOW_LOCAL_NETWORKS` now covers fake-IP proxies.
• Upgrade note: Any admitted owner can now use sandboxes; protect anonymous public deployments.

v1.4.1 (2026-11-05)

• Edits and learner progress preserved through background syncs.
• Background sync no longer discards unsaved edits, empties a PBL v2 project, or deletes a page an agent just added.
• Courses open quickly again on slow connections.
• PBL v3 releases a deleted course's sandbox.
• No upgrade conditions over 1.4.0.

v1.4.0 (2026-11-04) — PBL v3 Launch

• Project-based learning PBL v3: A new "vibe coding" pathway — build the project, understand how it works, learn by doing.
• AI tutor guides through requirements, design, implementation, and answers questions at any point.
• Milestones bring staged feedback and a personal learning report.
• Work and progress are kept for self-paced continuation.
• Enablement: Configure a sandbox provider and checkpoint storage, then open `/pbl-v3` (setup guide); stays off otherwise.
• One-way upgrade: Stop all 1.3.x instances first; rollback needs a database restore; older replicas exit when a newer release migrates the database. Read Breaking Changes before upgrading.

v1.3.x Highlights

• v1.3.1 (2026-11-02): DeepSeek Token Plan — one key lights up chat, web search, and web fetch; asset uploads no longer hold database connections across slow object stores; edits made just before leaving a page are kept.
• v1.3.0 (2026-10-29): Knowledge base and Pro workbench upgrade — every upload kept in owner's knowledge base until deleted, organized in folders, parsed once per conversation; composer offers it via `@`; agent can read, search, organize. Pro workbench shows agent work preview before first page exists; classroom tabs switch between Learn and Edit.

Deployment & Requirements

• PostgreSQL 16 required: Courses stored on server. For local development, `pnpm db:up` starts one in Docker.
• Server-first architecture (since v1.2.0): Generation runs on server and survives closed tabs/restarts; models configured once in `openmaic.yml`; web app and API share one pipeline.
• Long-running process needed: Course generation runs inside server and continues after browser leaves page. Vercel and other serverless hosts not supported from v1.2.0 onward.
• Docker Compose: Two containers (app + PostgreSQL). Courses, media, and runtime sessions live in named volumes `openmaic-postgres` and `openmaic-data`.
• Security: Set `ACCESS_CODE` (long random value, ≥16 chars) in `.env.local` to protect deployment. Set `PERSISTENCE_POSTGRES_PASSWORD` before first start.
• Optional services: Lemonade (local LLM/TTS/ASR), FunASR (local speech recognition), MinerU (advanced document parsing), VoxCPM2 (self-hosted TTS with voice cloning), render-service (MP4 video export).

Supported AI Providers

OpenAI, Azure OpenAI, Anthropic, Amazon Bedrock, Google Gemini, DeepSeek, Qwen, Kimi, MiniMax, Grok (xAI), OpenRouter, TokenDance, Doubao, Tencent Hunyuan/TokenHub, Xiaomi MiMo, GLM (Zhipu), Ollama, Lemonade, FunASR (local), and any OpenAI-compatible API.

Practical Implications

• Educators & creators: Build interactive courses without coding; export to PPTX/HTML for offline use.
• Developers: Integrate classroom generation into IDEs (Codex, DeepSeek) or messaging apps (Slack, Telegram, Feishu).
• Organizations: Self-host with full data control; configure token plans for multi-modal capabilities (chat, image, video, TTS, web search) with single keys.
• Verification: All releases documented on GitHub with changelogs, breaking changes, and upgrade guides. Source code at `github.com/THU-MAIC/OpenMAIC`.

How to Verify & Get Started

1. Visit the GitHub repository: `https://github.com/THU-MAIC/OpenMAIC`
2. Read the Quick Start guide and release notes for your target version.
3. For local development: clone repo, copy `.env.example` to `.env.local`, add `OPENAI_API_KEY` (or other provider key), run `pnpm db:up` then `pnpm dev`.
4. For Docker: copy `openmaic.example.yml` to `openmaic.yml`, configure providers, set `ACCESS_CODE` and `PERSISTENCE_POSTGRES_PASSWORD`, run `docker compose up -d --build`.
5. Review upgrade guides carefully — major versions (1.2.0, 1.3.0, 1.4.0) have one-way database migrations requiring full instance stops and database backups.