Editing eats hours. These three prompts hand you a rough cut, a color pass, and clean captions so you ship faster and keep your eye on the story.
[Claude]
You are a senior video editor. Given my transcript below and a target runtime of 90 seconds, produce a shot-by-shot edit plan as a table with columns: timecode in, timecode out, on-screen action, and the single line of VO or text that carries that moment. Cut every redundant sentence. Flag the one shot that must be a b-roll insert. Output only the table and a 3-line rationale for the biggest cut. Transcript: [PASTE]
[ChatGPT]
Act as a colorist. Given my footage is flat LOG from a Sony camera and the mood is warm, editorial, and interview-driven, write a DaVinci Resolve node tree for a primary grade plus one secondary node that isolates skin tones. For each node list the exact control, value, and why. Then give me a 5-shot LUT test checklist to confirm consistency across cuts. Output as a numbered node-by-node plan under 200 words.
[Cursor]
Write a Python script using ffmpeg-python that takes a folder of interview clips, detects silence with silencedetect, and outputs a CSV of every silent gap longer than 0.8s with start, end, and duration. Then generate a second CSV that proposes cut points that remove the longest silences while keeping each clip above 4 seconds. Print both CSVs to stdout and save them to ./cuts/. Handle filenames with spaces. No external deps beyond ffmpeg-python.
Full archive at aitoolalliance.com/tag/prompt-pack.