Orchard Core Videos

The source of every narrated Orchard Core video: the storyboards, slides, recorder scenes and YouTube metadata, and the skill that turns them into a video with one look and one voice.

The videos themselves are published on the Orchard Core YouTube channel and embedded from there in the documentation. No video file is ever committed, here or in the Orchard Core repository, so cloning a repository never downloads videos. Everything kept here is what's needed to make a video again, or to change how it appears on YouTube.

First time here? Follow SETUP.md: the one-time configuration of Google, YouTube and GitHub that lets the workflows upload videos and update the documentation.

Videos

Video	Length	YouTube	Source

This table is generated from each video's video.json by python tools/youtube/youtube.py index.

What's here

Path	What it is
videos/<slug>/	One video: video.json (YouTube title, description, tags, chapters, links, id), storyboard.json (narration), slides.py, recorder/Scenes*.cs, durations.json, captions, and a README on how to record it again
videos/*-walkthroughs/	Reels: existing silent screencasts joined into one video per area, built by scripts/reel.py from reel.json
.agents/skills/orchardcore-video-producer/	The skill: the video specification, the narration, recording and assembly scripts, the shared recorder, a worked example (examples/secrets-module/) and a template for a video's folder (templates/video/)
branding/	The Orchard Core logo and the Open Sans font every video uses
tools/youtube/	Uploads a video, pushes video.json changes, computes chapters, fills in the docs embeds, and writes the table above
.github/workflows/publish-video.yml	Publishing a release uploads its video and opens the docs pull request
.github/workflows/youtube-metadata.yml	Pushes a changed video.json to YouTube when it is merged
SETUP.md	The one-time setup, step by step: Google Cloud, the YouTube API, the repository secrets, and the docs token
CONTRIBUTING.md	How to submit a video, and how it is published

Making a video

To submit a video, read CONTRIBUTING.md: what a pull request holds, how a video is published to YouTube, and how the documentation pull request follows.

Follow .agents/skills/orchardcore-video-producer/SKILL.md. In short:

1. Copy .agents/skills/orchardcore-video-producer/templates/video/ to videos/<slug>/, and write the storyboard (start from examples/secrets-module/) and video.json.
2. Narrate (scripts/narrate.py), record the clips against a local Orchard Core site with fake data (the recorder), and assemble (scripts/assemble.py); then youtube.py chapters.
3. Check the video for private data, and merge its folder (never the MP4) to main.
4. python tools/youtube/youtube.py release <slug> --file <mp4> creates a draft release with the MP4 attached. Watch it there, then publish the release: the Publish video workflow uploads it to YouTube with its metadata, captions, thumbnail and playlists, records the YouTube id here, and opens a pull request in OrchardCMS/OrchardCore that fills in the <!-- video:<slug> --> markers of the docs pages.

A video's versions are its releases (<slug>/v1, <slug>/v2, ...): a re-recording is a new release, and the previous version on YouTube becomes unlisted. Changing a published video's title, description or chapters is an edit to its video.json; merging it updates YouTube. See SETUP.md and tools/youtube/README.md for the secrets these workflows need.

Requirements

Python 3.11+ (pip install pillow pygments imageio-ffmpeg "edge-tts>=7.3.1", and tools/youtube/requirements.txt to publish), the .NET 10 SDK for the recorder, and internet access for the voice.