AI Assistant Video Editing Brief Template
A clipping prompt runs one job; a brief defines how your video content always works. This template is the structured spec you attach to any video request so every assistant, and every teammate, produces clips that match your standards. It ends with a filled-in example for a 45-minute webinar.
The template
VIDEO EDITING BRIEF
Source: {video_url_or_upload}
Goal: {what these clips should achieve, e.g. drive signups, promote the episode}
Audience: {who should stop scrolling for this}
Platforms and formats: {e.g. TikTok and Reels, 9:16 vertical}
Tone: {e.g. direct and practical, no hype}
Clip length: {e.g. 20-60 seconds}
Clip count: {how many clips to produce}
Caption preset: {exact key from list_caption_presets}
Branding: {processing agent name, or logo/watermark rules}
Must include: {topics, announcements, or segments that must be covered}
Must avoid: {sponsor reads, confidential material, out-of-context claims}
Review criteria: {what makes a clip approvable, e.g. self-contained, accurate, strong first 3 seconds}
Approval flow: {who approves moments before rendering and clips before publishing}
----------------------------------------
EXAMPLE (filled in for a 45-minute product webinar):
Source: https://example.com/webinar-recording
Goal: drive trial signups for the new analytics feature
Audience: product managers at B2B SaaS companies
Platforms and formats: LinkedIn and YouTube Shorts, 9:16 vertical
Tone: confident and concrete, no buzzwords
Clip length: 30-60 seconds
Clip count: 4
Caption preset: pop
Branding: use the "webinar-brand" processing agent (logo bottom-right)
Must include: the live demo of the new dashboard, the pricing announcement
Must avoid: the Q&A about the delayed roadmap item, attendee names
Review criteria: each clip must make sense with no context, show the product visually at least once, and open on a question or claim
Approval flow: assistant proposes moments from list_clips, marketing lead approves before render_clip, final clips reviewed in OpenClip before schedulingFeatures
A spec, not a wish
Thirteen labeled fields turn "make some clips" into decisions an assistant can execute: formats, tone, exclusions, and what approvable means.
Team-proof by design
When the brief travels with the request, output stops depending on who briefed the assistant or which client they used.
Branding is a field, not a hope
Naming the processing agent (create_agent) in the brief pins captions, tracking, and watermark to the house style on every run.
The avoid list does real work
Sponsor reads, confidential slides, and out-of-context answers get excluded up front, which is much cheaper than catching them in review.
Review criteria you can enforce
"Self-contained, accurate, strong first 3 seconds" gives approvers a checklist and gives the assistant a target when recommending moments.
Approval flow made explicit
Who approves moments and who approves clips is written down, so automation speeds up the pipeline without blurring accountability.
Frequently Asked Questions
Write the brief once, reuse it forever
Copy the template, fill in your standards, and every assistant on the team clips to the same spec.