Video editing prompt library

Required inputs

• Podcast video or audio
• Target episode length
• Any moments that must stay

Tighten this podcast to about [TARGET LENGTH]. Remove only long, unintentional pauses and false starts. Keep thoughtful pauses, laughter, breaths that affect meaning, and every section marked [MUST-KEEP TIMECODES]. Do not rearrange the conversation. Add a short crossfade only where a cut clicks.

Required inputs

• Talking-head recording
• Filler words to target
• Terms or phrases to protect

Reduce repeated [FILLER WORDS] in this recording. Remove a filler only when the edit leaves a natural sentence and clean audio transition. Keep fillers used for emphasis or meaning. Never cut words in [PROTECTED PHRASES]. Preserve the original order and show me the cut list before finalizing.

Required inputs

• Full episode or selected section
• Audience
• Claims that need context

Find one 10–20 second exchange that gives [AUDIENCE] a clear reason to keep watching. Use only consecutive source dialogue; do not splice words into a new claim. Avoid any statement listed in [CONTEXT-SENSITIVE CLAIMS]. Place the exchange before the title, then return to the episode start. Provide the source timecode.

Required inputs

• Synced host and guest angles
• Wide shot if available
• Camera or speaker labels

Edit the synced podcast angles labeled [CAMERA LABELS]. Favor the active speaker, but hold useful listener reactions and use the wide shot to reset after long close-up runs. Do not switch cameras for less than 2 seconds unless hiding a cut. Keep all angles synchronized to the original audio and list any section without usable coverage.

Required inputs

• Episode or interview segment
• Desired duration
• Intended audience

Create a [DURATION]-second excerpt for [AUDIENCE]. Start with the question or the minimum setup needed, keep one complete idea, and end on the speaker's actual conclusion. You may remove repetition, but do not reorder sentences or combine non-adjacent words. Return the source ranges used and flag any context the short clip leaves out.

Required inputs

• Video with clear dialogue
• Language
• Names and spelling list

Caption this video in [LANGUAGE] using clean verbatim text. Keep the speaker's meaning and phrasing, omit non-meaningful stutters, and use the supplied spellings: [NAMES AND TERMS]. Use sentence case and punctuation. Keep each caption to two lines or fewer and break at natural phrase boundaries, not in the middle of a name or verb phrase.

Required inputs

• Captioned clip
• Brand accent color
• Words that must not be emphasized

Style the existing captions with [ACCENT COLOR]. Emphasize at most one meaningful word per caption phrase using weight or color, not both unless contrast requires it. Never emphasize [EXCLUDED WORDS], filler words, or punctuation. Keep all other words in a high-contrast neutral style and preserve the caption timing and wording.

Required inputs

• Multi-speaker clip
• Speaker names or neutral labels
• Preferred label format

Add speaker labels to captions only when the speaker changes and their identity is not visually obvious. Use [LABEL FORMAT] with these identities: [SPEAKER LIST]. Do not guess an uncertain identity; use [UNKNOWN SPEAKER] and flag the timecode for review. Keep labels separate from spoken text where possible.

Required inputs

• Video with final audio mix
• Audience needs
• Preferred cue style

Add concise non-speech cues in [CUE STYLE] only for sounds that change meaning, identify an off-screen event, or replace dialogue, such as [EXAMPLES]. Do not caption constant room tone or decorative music unless it carries narrative information. Place each cue at the sound and remove it when the sound ends.

Required inputs

• Video and current captions
• Approved glossary
• Language or locale

Proofread these [LOCALE] captions against the audio and [APPROVED GLOSSARY]. Report a timecoded correction list for transcription, spelling, punctuation, timing, and line-break errors. Do not rewrite for style unless text is inaccurate or unreadable. Mark uncertain words instead of guessing, and do not apply changes until I review the list.

Required inputs

• Landscape source
• Priority speaker or subject
• Caption placement

Reframe this landscape clip to 9:16. Keep [PRIORITY SUBJECT] in frame with comfortable headroom and protect the [CAPTION POSITION] caption area. Use slow, motivated reframing when the active speaker changes; do not chase small movements. If a crop would hide an important gesture or object, use a blurred fill and flag that section.

Required inputs

• Screen recording
• Presenter video
• UI areas that must remain visible

Place the presenter as picture-in-picture over this screen recording. Keep every area in [PROTECTED UI AREAS] unobstructed. Start at [PREFERRED CORNER], use consistent padding and size, and move the presenter only when a protected control would be covered. Animate moves briefly and keep the screen recording readable at the delivery resolution.

Required inputs

• Two synchronized speaker videos
• Output aspect ratio
• Speaker order

Build a [ASPECT RATIO] split screen with [SPEAKER ORDER]. Crop each source without stretching, match apparent head size, and align eye lines. Use the same divider and outer padding throughout. If one source cannot fill its half cleanly, use a neutral fill rather than enlarging it until facial detail breaks down.

Required inputs

• Product demo footage
• Important details or actions
• Maximum zoom preference

Add smooth punch-ins only when [IMPORTANT DETAILS] are discussed or demonstrated. Keep zoom at or below [MAXIMUM ZOOM], center the relevant product area, and hold long enough to inspect it. Ease into and out of each move. Do not scale beyond the point where the source becomes visibly soft; flag any detail that needs separate close-up footage.

Required inputs

• Final aspect ratio
• Required margin percentages
• Text and logo elements

Review this [ASPECT RATIO] edit using margins of [TOP], [RIGHT], [BOTTOM], and [LEFT]. Move [TEXT AND LOGO ELEMENTS] inside those guides without covering faces, hands, or demonstrated controls. Treat these as project planning margins, not official platform rules. Flag any frame where all constraints cannot be met.

Required inputs

• Final dialogue edit
• Speaker labels
• Delivery loudness requirement if known

Balance dialogue across [SPEAKER LABELS] so normal speech feels consistent. Reduce obvious level jumps, tame peaks without audible pumping, and preserve intentional whispers or emphasis. Use [DELIVERY REQUIREMENT] only if supplied; otherwise report measured loudness and true peak without claiming a platform target.

Required inputs

• Two or more camera angles
• Reference angle
• Known lighting changes

Match [OTHER ANGLES] to [REFERENCE ANGLE]. Correct exposure and neutral balance first, then bring skin tones and contrast closer without clipping highlights or crushing shadows. Preserve intentional lighting changes listed in [KNOWN CHANGES]. Apply corrections per shot and flag any angle that cannot be matched without visible artifacts.

Required inputs

• Edited narration or dialogue
• Approved B-roll library
• Claims requiring exact visuals

Add B-roll from [APPROVED LIBRARY] only where it shows the object, action, place, or evidence being discussed, or cleanly covers an existing edit. Keep the speaker visible for personal or emotional statements. Do not use a visually similar shot as proof of [EXACT CLAIMS]. Return a timecoded list pairing each insert with its source.

Required inputs

• Locked dialogue edit
• Licensed music track
• Sections that should be silent

Place [LICENSED TRACK] under the locked dialogue. Fade in and out around [SILENT SECTIONS], keep speech clearly dominant, and duck music smoothly under every voice. Let the track rise only in dialogue gaps or marked transitions. Do not retime the dialogue to the music, and avoid abrupt loop points.

Required inputs

• Near-final video
• Expected resolution and aspect ratio
• Required text or links

Review this near-final video for black or duplicate frames, flash frames, frozen footage, missing media, abrupt audio changes, caption errors, text clipping, and incorrect [REQUIRED TEXT OR LINKS]. Confirm the file is [EXPECTED RESOLUTION AND ASPECT RATIO]. Return a severity-ranked, timecoded issue list; do not silently change creative choices.

Required inputs

• Screen recording
• Required steps
• Maximum final duration

Tighten this tutorial to no more than [MAXIMUM DURATION]. Keep every action in [REQUIRED STEPS] visible. Shorten repetitive typing and loading, using a clearly visible speed-up or time jump rather than pretending the wait did not occur. Preserve enough lead-in and hold time for viewers to find each control.

Required inputs

• Screen recording
• Controls to highlight
• Brand color

Add a restrained [BRAND COLOR] cursor highlight and callout for each item in [CONTROLS]. Introduce the callout just before the action, keep its label concise, and remove it after the result is visible. Do not cover the target, menus, error messages, or entered values. Keep the real cursor path unless it is clearly accidental.

Required inputs

• Product recording
• Feature and user benefit
• Verified product claims

Create a [DURATION]-second reveal of [FEATURE] for [AUDIENCE]. Show the starting state, the real action, and the visible result in that order. Use only claims in [VERIFIED CLAIMS]. Keep UI labels readable, do not fabricate interactions, and end on the finished outcome with one concise benefit line.

Required inputs

• Unedited bug recording
• Expected behavior
• Sensitive information to hide

Edit this bug recording into a factual reproduction clip. Start with [EXPECTED BEHAVIOR], then show each action in order until the failure appears. Keep error text and relevant system state readable. Blur only [SENSITIVE INFORMATION], label any removed wait time, and add step numbers without covering controls. Do not infer the cause.

Required inputs

• Long tutorial recording
• Required chapter list
• Audience skill level

Structure this walkthrough for [AUDIENCE LEVEL] using these chapters: [CHAPTER LIST]. Open each chapter with a short outcome statement, then show the necessary actions in source order. Remove repetition across chapters but keep prerequisites before dependent steps. Add timecoded chapter markers and a final recap using only outcomes demonstrated on screen.