Generate, remix, poll, and download Sora videos via the OpenAI Video API.
npx skills add https://github.com/pchemguy/AISandbox --skill sora-pchemguy
Please help me install this Agent Skill. Skill: sora Source: https://github.com/pchemguy/AISandbox/tree/main/docs/AgentSkills/openai/skills/skills/.curated/sora Command: npx skills add https://github.com/pchemguy/AISandbox --skill sora-pchemguy
sora-2 or sora-2-pro, or remix an existing video by ID while keeping shot invariants. - Structured Prompt Augmentation: Automatically reformats prompts into a production-oriented spec (scene, camera, lighting, palette, constraints) for more controlled results. - Job Management & Batch Runs: Poll status, list jobs, download video/thumbnail/spritesheet variants, and run batch generation from a JSONL file. - Use Case: You need a 4-second product teaser of a matte black camera for a landing page. The Skill augments your prompt, submits the job, polls until complete, and downloads the MP4 locally. ## Quick Start Ask the AI to generate a 4-second Sora video of your subject using the bundled CLI, then poll and download the finished MP4 to your project folder.High-intent search queries and answers about installing and using this skill.
Run the bundled CLI with `create` or `create-and-poll`, passing a prompt, model, size, and seconds. The CLI submits the job to the OpenAI Video API, and create-and-poll waits for completion and optionally downloads the MP4.
Use the `remix` command with the video ID and a prompt describing the change. State invariants explicitly, such as same shot and camera move, and change only one element per remix to avoid drift.
Models are limited to sora-2 and sora-2-pro. Durations must be 4, 8, or 12 seconds passed as a string. sora-2 supports 1280x720 and 720x1280; sora-2-pro adds 1024x1792 and 1792x1024.
The size must match the selected model and seconds must be one of 4, 8, or 12 as a string value. Check references/video-api.md for the supported size list per model and pass seconds via the API parameter, not in the prompt text.
Yes, pass a jpg, png, or webp image via --input-reference to anchor the first frame. The image should match the target size, and images containing human faces are rejected by the API.
Download URLs expire after about one hour, so re-download while the link is fresh and copy assets to your own storage. Also confirm the job status is completed, since downloads are only available after completion.
cfongaiConvert comic panels, manga pages, and storyboards into animated videos using Seedance 2.0.
MoizIbnYousafRecord product demos and walkthrough videos using ply screenshots, ffmpeg stitching, or Remotion compositions.
VictaaawGenerate marketing videos with AI avatars, voiceovers, and programmatic templates.
eachlabsAdjust video playback speed from 0.25x to 20x with frame interpolation.