Skip to content
cccskills
FreePublished on GitHub

footage-cutlist

Watch a brand's own product footage (screen recording, product film), pick the moment that proves each line with the user, and render those picks as a product layer — free, local. survey.py turns footage into timestamped contact sheets + scene cuts + motion so the agent can actually look at it; the agent writes a cut list (beat → source window + framing); preview.py draws it as a review sheet the user corrects round by round; cut.py renders the approved list into a silent 1080x1920 layer with the creator's area left as a plate. Also frames footage as a filmed screen (keystone, bezel, handheld drift) with dissolves between beats, and takes still screenshots as sources. Use for any format that shows real product footage beside, between or instead of a creator (split-screen, screen inserts, walkthroughs).

The English summary is being prepared. Showing the original description.

Run score Not tested

See how to install

Files included (10)

  • SKILL.md8.0 KB
  • scripts/_common.py4.2 KB
  • scripts/cut.py11.6 KB
  • scripts/cutlist.py7.7 KB
  • scripts/filmed.py5.7 KB
  • scripts/photo.py2.7 KB
  • scripts/preview.py7.2 KB
  • scripts/survey.py6.1 KB
  • skill.meta.json304 B
  • tests/smoke-test.md853 B

SKILL.md (original)

Review the instructions the agent will receive before you install.

footage-cutlist

Picking which moment of a customer's footage plays under which line is a judgement, not a formula. A position map ("beat 3 is 40% through the reel, so take the footage 40% through") was measured up to 5.7s off a person's picks on a reference build. So this capability does the free, mechanical parts and leaves the decision to the agent and the user:

StepScriptWho decides
Look at the footagesurvey.pynobody; it just shows it
Choose a window per linethe agent writes cutlist.jsonagent, then the user
Show the choicespreview.pythe user corrects them
Render the approved choicescut.pynobody

All free. ffmpeg + Pillow + numpy only.

Run

python survey.py --video footage.mp4 --out work/survey            # whole clip, ~1 frame/s
python survey.py --video footage.mp4 --out work/s-12 --from 12 --to 18 --every 0.25 --frames
python preview.py --cutlist work/cutlist.json --out work/review.png
python cut.py --cutlist work/cutlist.json --out work/layer.mp4 [--draft]

survey.json lists cuts (scene changes), motion (change per second) and still (runs of 1.5s+ with nothing moving: a dead screen reads as a frozen video). Read the sheets; open a single frame-*.png (from --frames) when you need to read small UI text.

The cut list

{
  "size": [1080, 1920], "fps": 30, "seam": 768, "creator_side": "bottom", "bg": "auto",
  "sources": {"demo": "footage/demo.mp4"},
  "beats": [
    {"id": "b1", "start": 0.0, "end": 3.1, "state": "creator", "vo": "..."},
    {"id": "b2", "start": 3.1, "end": 7.4, "state": "split", "vo": "...",
     "source": "demo", "in": 12.0, "out": 16.3, "fit": "width", "why": "the brand kit fills in at 13.4s"}
  ]
}
  • state: creator (creator full frame), split (footage in the product zone, creator in the other), product (footage full frame).
  • in is where the window starts in the source; out defaults to in + slot (1.0x). A window may play between 0.5x and 2x; it is never looped or frozen to fill a slot.
  • fit: width (whole frame, letterboxed; the default), cover (fill + crop around focus), crop (a source box [x0,y0,x1,y1] in fractions, then fit by width).
  • bg: auto samples the footage's own corner colour so the letterbox and the footage read as one surface; blur; or #rrggbb.
  • look: "screen": frame the footage as a screen filmed at close range (thin bezel, dark room, faint moire, grain, slow handheld drift). crop picks the source region, screen: {rot, keystone, fill, fit, drift} the geometry, mask boxes are blurred in the source (an email, a customer name). Inserts over a creator: keystone ~0.007, rot within ±1°. Walkthroughs: keystone ~0.024, rot -1.5..0, fill ~0.80. Never bigger angles and never alternate them between beats (read as a wonky camera).
  • transition: {"dissolve_frames": 7} (top level): cross-dissolve every beat into the next. A hard cut between two screens reads as an edit; 6-7 frames reads as the camera moving.
  • A source can be a still image (png/jpg/webp): it holds for its slot. Give it look: "screen" so the drift keeps it alive.
  • why: one line on what the window shows. It is printed on the review sheet so the user can see the reasoning, and it makes the agent say what it saw.

Beats must tile the timeline from 0 with no gaps. cut.py refuses a list that does not.

Rules that were learned the hard way

  1. Watch the whole thing before choosing anything. Look at every sheet of the whole-clip survey first. The moment that proves a line is often not where the recording's order suggests.
  2. Never commit a window from a 1fps glance. Re-survey that range at --every 0.25 and look at every frame. A click, a page load or a half-typed field lives between the samples, and it is exactly what a user then points out ("you didn't look at the video properly").
  3. The line must be true of the footage while it is said. If a line says "it pulls your brand colours", the colours must be on screen during that line, not a second later. Name what is visible in why.
  4. Never crop footage that has copy in it. Cropping a UI panel into a narrower zone slices through words. Default to fit: width; use crop only to isolate a small subject, and check the review sheet for cut-off text.
  5. Avoid dead screens. A window inside a still run reads as the video having stopped. Start it where something moves, or shorten it.
  6. A screen region should be roughly the box's shape. A 9:16 box is filled by a source region of ~0.6-0.8 aspect: crop fewer COLUMNS at full height. A wide, short crop floats in black. For a complete panel with copy, use fit: width and let the room fill.
  7. Watch for a third party inside the footage. A customer logo row or another company's name in the recording ends up in the ad. Crop above it or mask it.
  8. Don't reuse footage. cut.py warns when two beats overlap in the same source: it reads as a loop.
  9. After the takes exist, re-check. Line timing moves when the real voice is aligned (align_beats.py). Re-run preview.py + cut.py on the aligned list and look again.

Going back and forth with the user

Expect several rounds. Each round:

  1. Show review.png (and a --draft layer if they want to see motion).
  2. The user names beats: "b4 is wrong, use the part where the report loads."
  3. Re-survey just that stretch densely, find the moment, and change only that beat's window. Say what you changed and why.
  4. Re-render the review sheet and show it again.

Picking is free, so take as many rounds as the user needs. Never spend on the creator until the user has approved the cut list.

Source paths resolve relative to the cut-list file, never the shell working directory. Use absolute source paths when preparing a run. Save explicit inspected in/out bounds; defaults are compatibility behavior, not evidence that a window was watched. Screen framing defaults to width to preserve UI text; height fitting is an explicit reviewed choice.

Product-photo motion and caption review

A real product photo can use look: "photo". This is a dedicated contain/pan/zoom treatment without a screen bezel. photo accepts start/end scale (0.85–1.0), start/end pan pairs (-1–1), a normalized content box, and normalized protected source rectangles. An optional crop must contain every protected product/text rectangle. The selected region remains visible throughout motion. Use photo.box to reserve space above/below the seam for a qualification. Review first/middle/last frames; preview and render share the photo framing function. Existing plain stills and screen looks remain supported.

Before showing the cutlist, generate a footprint with caption-burn's bundled footprint.py, using the final style, anchor and font. Pass that JSON to preview's caption-footprint option. The amber band is the union of the actual rendered caption groups for that beat. Rebuild it after copy/timing/style changes. If it covers qualifying text, adjust photo.box, crop or layout, then inspect the final captioned video. Do not reduce or hide the qualification to make the claim fit.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Research supported advertising promises from a specific buyer's context, current product facts, customer language and observed ads. Turn them into static, video or copy ideas; match video ideas to actual recipes and save a product-scoped angle bank for write-video-ad-script. Product-led discovery can use verified facts without reviews. Reference persistence and reach are observations, not proof of performance.

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

Analyze ad campaign performance data (Google, Meta, LinkedIn) to identify what's working, what's wasting budget, and specific cut/scale/test recommendations. Runs statistical analysis, funnel diagnostics, and multi-channel budget reallocation with specific dollar-amount shift recommendations and scenario modeling.

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

For paid lead-gen and participant-recruitment ads, replaces vanity CPA with true CAC per qualified lead by joining ad-platform data with downstream funnel events, surfaces tracking gaps, and classifies every creative into Scale / Keep / Investigate / Cut.

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

The intake stage of a Meta ad harness. Researches the brand first, states what it assumes, and asks only what it cannot infer and what would change money, claims or legal exposure if guessed wrong — at most four questions a turn, in plain business language. Writes the brand's stable facts (ads/brand.md), the campaign brief, and a log of what was asked vs assumed. Use it when someone wants to run Meta ads and the brand's ad facts are not yet recorded.

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

Checks paid ads against the ad platform's rules (Meta, TikTok, Google, LinkedIn and others) and the brand's category rules: the copy, on-image or on-screen text, pictures and people. Rewrites what can be fixed honestly, adds the disclosures that are needed, and flags anything that needs a human or legal review, with one pass, fixed or flag line per ad and platform. Use alone when you already have the ad text or images and know where they will run; otherwise run the playbook for the job (e.g. make-static-ads, make-content, email-lifecycle).

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

Analyze the message match between your ads and landing pages. Checks if the promise in the ad copy carries through to the landing page headline, body, and CTA. Flags disconnects that kill conversion rates. Works with Google, Meta, and LinkedIn ads.

The English summary is being prepared. Showing the original description.

gooseworks-ai/goose-skills1.3KUpdated Oct 24, 2026

See all skills by gooseworks-ai

Report a problem with this skill