waoowaoo's preview: an install that is a prompt, an Elastic license, and a compose file that fails closed
Industry-first professional AI agent platform for controllable film and video production, from shorts to live-action, with Hollywood-standard workflows.
At a glance
- What is it?
- waoowaoo is a TypeScript AI workspace for image and video generation that ships as a self-hosted Docker preview on top of Next.js 16, Temporal, MySQL and MinIO, with its quick start written as a prompt to paste into a coding agent.
- Who is it for?
- waoowaoo fits a team that wants a self-hosted canvas for AI image and video work, is willing to supply its own OpenRouter key and Docker host, and accepts a preview surface with no music or voiceover. It does not fit anyone planning to offer it as a hosted service to third parties, anyone expecting open source licensing, or anyone who wants their pull requests merged.
- Can I use it commercially?
- Check first. The repository uses a licence we do not classify automatically, so read its LICENSE file before any commercial use.
- Is it still maintained?
- Yes. The repository last received commits 55 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- waooAI/waoowaoo14,866 ★ · +833
| Repository | 2026-09-10 | 2026-11-15 |
|---|---|---|
| waooAI/waoowaoo | 14k | 14.9k |
833 Stars in the last 66 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
The quick start is a prompt to paste into a coding agent, not a command list
The installation section does not give you shell commands. It gives you a block of text intended to be copied into a local coding assistant with terminal access, and it explains that the assistant will inspect the machine, install what is missing, and configure Docker. The text block itself spells out the contract:
Install the self-hosted preview of https://github.com/waooAI/waoowaoo on my computer.
Read the repository README, docs/INSTALL.md, and the selected GitHub Release first.
Use a published preview release and its prebuilt Docker images. Pin both the
application and the Codex runtime to the release's immutable image digests.
Check my OS, CPU architecture, Docker Engine/Desktop, Compose v2, Git, and flock;
install missing dependencies using the appropriate OS tools, with approval where required.
Use a new installation directory and preserve all existing containers, files, and data.
GenerateThe project also warns that you may need to approve an operating system installer or start Docker Desktop yourself, so this path assumes an interactive machine with a human in the loop.
The license switched at the preview tag and it is not open source
The license section is the most consequential paragraph in the file. Starting with v0.5.0-beta.1, the distribution is licensed under the Elastic License 2.0, and the project states plainly that this is source-available software, not OSI approved open source. What is permitted: personal use, internal business use, modification, and redistribution, subject to the license terms. What is not: providing third parties with hosted or managed access to a substantial set of the software's features, which requires separate permission. That boundary defines who may run it. A studio using it internally is inside the grant; a vendor offering multi tenant access to customers is outside it and has to ask. The same paragraph notes that earlier releases retain their original licenses and that third party components keep their own, so the terms you get depend on which tag you deploy.
External pull requests are read for reference and implemented internally
The contributing section is unusually direct. Bugs and feature requests go to the issue tracker, and you may submit a pull request for reference, but the project adds that the team reviews ideas carefully and implements changes internally rather than merging external PRs directly. Read that as a workflow fact rather than a courtesy: your branch will not land in the release line, and time spent building a patch against a preview the team says it is iterating on rapidly has no guaranteed payoff. The project is maintained by the core team, and the file positions itself as a preview release where the team says it is iterating rapidly and some bugs and rough edges remain. The two statements together describe a codebase that moves under you, so the useful contribution is a well described issue with a reproduction, and the useful check is whether a fix already landed upstream before you build on a tag.
The preview has no audio controls, and one model takes a single frame only
Scope is narrower than the project description suggests, and the file is candid about where. The preview exposes OpenRouter configuration, with available models including GPT Image 2, Nano Banana variants, Seedance variants, and MiniMax H3 and H3 Max, and it notes that availability and charges depend on the provider. Two specific gaps are called out. Music and voiceover controls are not included in this preview, so a finished film needs audio handled outside the tool. And H3 Max currently accepts a single first frame in this application, not a first and last frame pair, which means the first/last-frame mode listed in the features is unavailable for that model even though the mode itself is described as supported when the selected model allows it. Everything else in the feature list is generation shaping: model aware aspect ratios, durations, resolutions, and reference roles, plus refining a result into a new version while keeping the original.
Media stays local, but prompts and references are sent to the provider
The API configuration section draws a line that matters for anyone with unreleased footage. Project data and media use a local database and private MinIO storage, so your files sit in your own infrastructure rather than a vendor's bucket. Supported inline image references do not require a public media URL, which means you can reference your own assets without exposing them to the open internet. However, prompts and references are sent to the provider for requested AI tasks, so the content of what you submit crosses that boundary even though the stored copy does not. Billing follows the same line: you add your own OpenRouter API key in the profile's API configuration, provider calls are charged to your own account, and the project asks that keys be kept out of chat and support logs. There is no bundled credit and no bundled provider, so cost scales with your own usage and cannot be estimated from anything in the repository.
The compose file refuses to start until every secret is filled in
Environment handling is strict by design, and reading the compose file shows how. DATABASE_URL, REDIS_PASSWORD, NEXTAUTH_SECRET, CRON_SECRET, and API_ENCRYPTION_KEY are all interpolated with a required marker, so an unset value stops the stack instead of booting with an empty secret. Other variables carry defaults instead, with S3_REGION falling back to `us-east-1` and storage endpoints defaulting to empty because edition overlays own the concrete endpoint, while the base file stays composable. A comment states that Temporal owns separate schemas and never uses the application database URL, which is the kind of detail that saves an hour if you plan a backup. The environment example then hands you an APP_IMAGE digest made of all zeros with a comment telling you to replace it with the same repository and sha256 digest used by both Worker slots. Copy that file as it stands and the pull fails on a placeholder.
Two port schemes and four compose files, so pick the one you are running
The tree carries docker-compose.yml plus dev, self-hosted, self-hosted dev, and test overlays, and the scripts show how they combine. The infrastructure line names the services it starts, running `docker compose -f docker-compose.yml -f docker-compose.self-hosted.yml up -d` against minio, minio-init, mysql, redis, and a temporal-create-namespace target. Two different port schemes follow from which path you take. The release Compose setup starts Caddy automatically, and after you approve local certificate trust you open `https://localhost:1443`, while port `13000` only redirects HTTP. Source development with `npm run dev` stays on `http://localhost:3001`, and the environment file carries both sets along with the note that NEXTAUTH_URL is derived from the host and HTTPS port in Docker mode. The base image is node:22-bookworm-slim rather than Alpine, because Prisma has to detect the same OpenSSL ABI it will use at runtime and Temporal's native bridge needs glibc.
Editorial conclusion
waoowaoo fits a team that wants a self-hosted canvas for AI image and video work, is willing to supply its own OpenRouter key and Docker host, and accepts a preview surface with no music or voiceover. It does not fit anyone planning to offer it as a hosted service to third parties, anyone expecting open source licensing, or anyone who wants their pull requests merged. Before installing, read docs/INSTALL.md, replace the all zero image digest in your environment file with the digest the release publishes for both worker slots, and generate every secret the compose file marks as required.
Frequently asked questions
What does waoowaoo actually do in this preview?
It is an AI creative workspace for images and videos where the Assistant develops a brief, you upload reference material, and generated results are organized and refined on a canvas. The preview exposes OpenRouter with GPT Image 2, Nano Banana variants, Seedance variants, and MiniMax H3 and H3 Max.
Can I run waoowaoo for clients or as a hosted service?
Not without asking. From v0.5.0-beta.1 the distribution is under the Elastic License 2.0, which permits personal use, internal business use, modification, and redistribution, but giving third parties hosted or managed access to a substantial set of features requires separate permission. The project states it is source-available, not OSI approved open source.
What will a waoowaoo installation cost me?
You bring your own OpenRouter API key in the profile's API configuration and provider calls are charged to your account, with availability and charges set by the provider. The application itself runs on prebuilt Docker images, so you need no Node.js, MySQL, Redis, Temporal, or FFmpeg on the host.
Does waoowaoo do music and voiceover yet?
No, music and voiceover controls are not included in this preview. Generation is limited to the supported modes of the selected model, and H3 Max accepts a single first frame in this application rather than a first and last frame pair.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/waooai-waoowaoo)Star history chart for your README
[](https://hysenlabs.com/projects/waooai-waoowaoo)Was this useful?