zhenxin-ai-video-manager

You have a pile of raw footage and want a finished video, but the idea of dragging clips on a timeline, adjusting subtitles, and mixing voiceover in a NLE feels like a chore. This plugin takes a different route: describe what you want in plain language, and the agent orchestrates the entire pipeline — material analysis, rough cut, subtitle burn-in, local AI voiceover, and final composite — using open-source tools that all run on your own machine. Nothing leaves your computer.

It follows a conservative editing philosophy: only clearly redundant gaps, duplicates, and bloopers are trimmed automatically; anything semantically uncertain is kept as-is. Source files are read-only — never moved, deleted, or overwritten. Every revision produces a new version, and any clip you remove can be recovered from the original material at any time. Before handing the result over, it verifies the output with FFprobe to confirm the file actually exists, is readable, and matches the target aspect ratio and duration.

It is built for solo creators, podcasters, and instructional-video makers who have real footage and need a fast turnaround without becoming editors. It does not generate footage from scratch, does not auto-publish to any platform, and does not call paid cloud services. If you want a second pass in CapCut or Jianying, an optional draft export lets you open the project and keep polishing right where the agent left off.