wtf is book-video-factory?

Book Video Factory is an installable skill for Codex that helps you produce short review videos for Chinese books, while keeping a careful record of rights, evidence, and approvals along the way. It supports two different video styles: the original Bilingual Editorial Book Video, and a newer VOX-style Book Video that leans on voice narration paired with short visual scenes. Importantly, the repository itself does not include any book content, covers, music, sound effects, voice samples, or credentials. Everything shown in the examples folder is a finished showcase video, not raw material bundled with the tool, and those demo videos are not covered by the project's MIT license, since books, music, and other elements belong to their original rights holders. You install the skill through a command line tool called npx skills, then open a new task in Codex and ask it to bootstrap a project for a specific book title and author. From there the skill sets up a consistent folder structure, a bilingual script template, and links each line of the script back to its supporting claims and source material. Every decision along the way, including technical checks, story review, background music rights, and final publication, has to be approved separately by a human, and nothing gets marked as ready to release unless the file hashes and the recorded approvals actually match up. Costs from any AI providers used are logged honestly, without inventing numbers. For the newer VOX-style videos, there are two ways to generate visuals: through Google's Gemini API for programmatic, repeatable generation, or through Google Flow's manual creative interface for a human to adjust each scene by hand. Either way, the tool records the prompt, the model used, input and output file hashes, and who approved each step, and any AI-generated audio gets removed unless it has its own separate approval. The overall workflow moves through locking down an approved script, generating narration and timing, approving the visual concept and layout in stages, generating the actual clips, and finally combining picture, narration, captions, music, and branding into a finished video that passes a technical check before publication.