Directs the creative quality of a video before detailed production. Use when Codex needs to define style, reference direction, audience fit, hook strength, narrative angle, visual intention, pacing target, platform tone, or decide what a video should feel like before script, assets, edit plan, or Remotion implementation.
Makes high-level editing decisions for a polished video. Use after script and footage/asset analysis to choose the opening shot, pacing, cuts, shot order, J-cuts/L-cuts, cutaways, montage rhythm, retention beats, and what to remove or emphasize before implementation.
Analyzes raw footage before editing. Use when Codex receives existing video files and needs to extract frames, detect shots, identify usable moments, locate text or logos, score visual quality, flag bad footage, and create a shot list before script, timeline, or edit decisions.
Designs the visual finish of a video. Use when Codex needs to improve perceived production quality through color style, typography, title design, caption style, layout rhythm, brand palette, LUT notes, contrast, grain, sharpness, and platform-safe visual polish.
Reviews finished or draft videos for quality, not just technical correctness. Use after preview or render to score hook strength, clarity, pacing, visual polish, audio, captions, platform fit, originality, and production value, then route fixes back to the responsible skill.
Routes multi-step video production requests across specialized video skills. Use at the start of any video creation, editing, Remotion, asset, voiceover, audio, motion, packaging, or export task to decide which skill handles the current layer and what artifact should be produced next.
Finds, generates, downloads, validates, transcodes, and manifests video production assets. Use when Codex needs to automatically obtain video clips, images, screenshots, generated visuals, icons, data media, music candidates, or sound effects for a video. Do not use for final timeline assembly or aesthetic editing.
Plans how gathered assets are used in the video timeline. Use to assign assets to spoken beats, de-duplicate clips/images, map A-roll/B-roll, create edit plans, align assets with narration timing, and decide scene-level asset usage. Do not use for downloading assets.