Polish a screen recording

What it asks you for

Recordingrequired

The raw capture.

Title card

Optional opening title.

Cut dead air

Remove long silent gaps.

What it tells the AI to do

The full instructions, exactly as they run.

Polish {{ video }}.

- Title card: {{ title }}
- Cut dead air: {{ removeSilence | default(true) }}

A raw screencast is mostly waiting. Detect silences with ffmpeg's
`silencedetect`, then cut gaps longer than ~0.8s — shorter gaps are natural
speech rhythm and cutting them sounds robotic.

Add the title card as a 2s clip concatenated at the front (generate it as an
image and burn the text in code, not with a diffusion model). Normalize audio to
about −16 LUFS so it matches everything else the user will publish.

Deliver the polished file AND report how much time was removed — that number is
how the user judges whether the cut was too aggressive.
## Delivery

Write the finished artifact into Drive. Never upload a user's content to a
third-party file host (`0x0.st`, `catbox.moe`, `transfer.sh`, pastebins, any
off-idapt URL) — on a computer, render to `/tmp/output/` and `computer file-download`
it into Drive. If you catch yourself reaching for `curl -F`, stop: you already
hold the artifact.

• video
• screencast
• demo

More video skills

• Create a videoGenerate a video clip from a text prompt.
• Animate an imageTurn a still image into a short moving clip.
• Edit a videoTrim, join, speed, crop, or watermark a video with ffmpeg.