Your AI agent can publish a video track by streaming a

video/mp4

file using the LiveKit SDKs. Here are the main approaches: • Python SDK: Use the

av

library to decode the MP4 file, convert frames to the required format, and yield them as

VideoFrame

objects. Create a

VideoSource

and

LocalVideoTrack

, then publish the track to the room. This method allows direct streaming of MP4 files. • Go SDK: First, convert your MP4 file to a supported format (like H.264) using

ffmpeg

. Then, use

NewLocalFileTrack

to publish the encoded file as a video track. • LiveKit CLI: Convert the MP4 to H.264, then use the CLI to join a room and publish the video file as a track. Key requirements: • MP4 files must be converted to supported codecs (H.264, VP8, or VP9) before publishing • Set the correct frame rate and video dimensions when publishing If you need a step-by-step code example for a specific SDK (Python or Go), let me know which one you prefer. Sources: examples/video-stream/video_play.py | examples/filesender/main.go | cmd/lk/join.go