Pinokio launcher for OpenMagia, a local-first AI video editor. The launcher repository contains only lifecycle scripts; it clones the application into app/, leaving the main OpenMagia repository independently maintained.
Paste this repository's public GitHub URL into Pinokio's Download from URL flow. Until clean-machine validation is complete, this launcher should be described as a release candidate rather than advertised as verified one-click support.
- Install Editor clones OpenMagia and validates its Python sources and regression suite without downloading model weights.
- Start runs
server.pyas a Pinokio-owned daemon on a dynamically allocated port. - Open UI appears when the server is ready.
- Stop stops the Pinokio daemon.
- Update fast-forwards both launcher and application repositories, then runs OpenMagia's regression suite before reporting success.
- Install Generation Model is a separate, explicit, license-gated download on supported Macs.
Updates do not delete or reset projects/, media/, uploads/, data/, models/, or addons/. These paths are ignored by the application repository and remain inside Pinokio's managed launcher directory.
The local compatibility checkout is validated against OpenMagia 89c1289: shared Qwen Image model management, H3MLX support, completed-character queue recovery, the in-app New Project dialog, compact audio/video asset previews, preserved storyboard audio settings, and local AI continuity repair previews. The launcher keeps one application source in app/; it does not maintain a second copy of the editor implementation.
Public Install and Update follow OpenMagia’s upstream main. Local commits must be published to that branch before existing public installs can receive them. This local compatibility review does not constitute a fresh Pinokio GUI installation or new cross-platform verification.
| Capability | macOS Apple Silicon | macOS Intel | Linux | Windows |
|---|---|---|---|---|
| Editor and existing-media workflow | Supported | Expected; clean-machine test pending | Expected; clean-machine test pending | Expected; clean-machine test pending |
| Timeline export | Supported with FFmpeg | Requires FFmpeg; test pending | Requires FFmpeg; test pending | Requires FFmpeg; test pending |
| Native MiniMax H3 generation | Supported with 64+ GB RAM and ~278 GB disk | Not supported | Not supported (CUDA backend not yet integrated) | Not supported (CUDA backend not yet integrated) |
The generation action requires the user to review and explicitly accept the MiniMax H3 Community License. It invokes OpenMagia's idempotent installer, so interrupted Hugging Face downloads resume rather than restart.
For automated testing after the operator has accepted that license, the same action can be invoked through Pinokio's terminal client with an explicit argument:
pterm start install-model.js --ref pinokio://127.0.0.1:42000/api/OpenMagia-Pinokio -- --acceptance='I ACCEPT'Any other value still opens the license prompt and the installer independently checks the exact acceptance text before downloading.
Once Start reports that OpenMagia is ready, use the URL shown by Pinokio (for example, http://localhost:42003). The API is local and requires no cloud account.
JavaScript:
const state = await fetch("http://localhost:42003/api/state").then(response => response.json())
console.log(state.engine)Python:
import requests
state = requests.get("http://localhost:42003/api/state", timeout=10).json()
print(state["engine"])cURL:
curl --fail --show-error http://localhost:42003/api/stateReplace 42003 with the dynamic port displayed by Pinokio. Project creation is available via POST /api/projects with JSON such as {"name":"My project"}, and media can be imported with POST /api/upload plus an X-File-Name header and the file bytes as the request body.
See TESTING.md. A registry listing does not imply endorsement or verified compatibility. Verified Discover placement requires publisher verification, review/testing, and transfer of the launcher repository to Pinokio Factory under Pinokio's current publishing policy.
The launcher scripts are available under the MIT License. OpenMagia and downloaded models retain their own licenses.
