CapCut Draft Assistant

Problem

A common pain point in CapCut workflows is not single-shot editing, but assembling scattered images, videos, audio, subtitles, and copy into one consistent timeline. Manual imports, timing, stickers, and draft saving can become repetitive, while local automation scripts often require deployment and API keys.

How it works

CapCut Draft Assistant targets editable draft generation. It first clarifies canvas ratio, material order, duration, subtitle mapping, and public material URLs from natural-language requests. It then calls create_draft, builds timelines with timelines or audio_timelines, uses *_infos to produce structured JSON strings, writes materials and effects through add_* endpoints, and finishes with save_draft to return a draft_url. Sticker and animation selection should use search_sticker and verify title matches; keyframe offsets are segment-relative microseconds.

Boundaries

It delivers editable CapCut drafts, not cloud-rendered final videos. Local installation targets the Windows CapCut draft root, defaults to refusing overwrite of same-name drafts, and expects accessible public URLs. Do not submit cookies, tokens, captchas, or enterprise-sensitive data.