OrkasVideoStudio
A tool that lets a coding agent compose, generate and edit video from a plain-language description, expressing the result as an editable, re-renderable plan rather than a black-box render.
OrkasVideoStudio represents a video as a readable, diffable plan.json that an agent or a person can edit; changing one line re-renders only that piece. It wraps hyperframes, ffmpeg and whisper.cpp as deterministic capabilities for rendering, editing, transcribing and generating video, alongside knowledge of what makes a good video and which production line to use for it.
Three axes cover most work: Compose turns a script into designed HTML motion graphics such as explainers, kinetic typography, lower-thirds and title cards; Edit cuts, joins, trims silence, removes filler words, mixes audio, burns in subtitles, dubs and localizes supplied footage; and Generate produces talking-head or cinematic AI footage and imagery through a user's own OpenAI, Gemini or Doubao provider keys. Compose and Edit need no paid keys.
When a deliverable needs more than one axis, an automatic end-to-end pipeline builds an edit decision list, assembles it by delegating each segment to compose, generate or edit, and runs a deterministic delivery guard that checks the finished cut before it ships. It works from any coding agent that can run a shell or speak MCP, and began as the built-in video agent inside a separate desktop application, Orkas.