dsh-video-studio

Most video editing tools live outside the AI conversation, forcing you to bounce between a chat window for scripting and a separate editor for cutting. dsh-video-studio collapses that split by embedding a full four-lane timeline—video, audio, titles, and captions—directly inside a DeepSeek Harness conversation, so you can debate a concept with the model and immediately drag, trim, and reframe clips on the canvas right beside it.

The plugin anchors on three layers of capability. First, Remotion-driven, frame-accurate preview and MP4 export stay locked to the same composition frame that GSAP animations use, giving you genuine WYSIWYG confidence before you hit render. Second, a complete editing pipeline: pull media from your workspace into a managed library, arrange and split clips on the timeline, adjust volume and framing, author or round-trip SRT captions across portrait and square canvases, and undo or redo with full confidence. Third, a tight collaboration contract with DSH itself—one click hands the current project, selected clip, and playhead back to the conversation draft, while the Agent can read, update, or queue a render through versioned tool calls, with automatic local backups on conflict.

It is built for creators and small teams who produce short-form video, demo clips, or captioned presentations inside an AI-assisted workflow and want the distance between writing and editing to nearly vanish. The compositing model favors fixed lane types and a coherent, lightweight in-conversation editor over multi-cam, keyframe-curve, or real-time collaboration features, positioning the plugin as a focused video workbench rather than a full NLE.