TikTok creator uses Claude Code to automate video editing and scripting

Brooke Lacey, a TikTok creator with over 200,000 followers, has developed a workflow using Claude Code to automate the production of short-form vertical video. The system operates locally on a MacBook, utilizing existing tools such as FFmpeg, Whisper, and Python without uploading video files to external servers. The editing process involves silence detection to mark speech, word timing from Whisper to sharpen cut edges, and facial tracking to adjust crop and shot size for each cut. The edit is defined by a text-based TSV file containing start and end times, shot sizes, and labels, allowing the same command to generate 15-second, 30-second, 60-second, and landscape versions from a single take. Lacey notes that the system reduced editing time from hours to approximately 15 seconds of processing for a 5-minute raw take, producing videos with precise jump cuts and TikTok-appropriate cadence. The agent also generates captions and inserts app screenshots or graphics at specific frames.

Lacey also uses Claude Code to analyze her existing TikTok content corpus to generate new video scripts. Because TikTok’s backend prevents direct scraping, she manually downloads videos via her phone and uploads them to Google Drive, where Claude Code can access them. The agent pairs these video files with live comment threads scraped from the TikTok interface to identify audience reactions and topics. For example, it analyzed comments on a video about AI in tech to generate new content ideas based on user queries, such as AI in hospital billing. Lacey has trained the agent on her specific speaking cadence, facial expressions, and hook styles by referencing her past videos, enabling the AI to write scripts that mimic her unique voice and comedic timing. This workflow supports her production of ads for her app, Zosia, a free AI learning application.