agentleFS
konradrekonradre/video-production-skills/skills/video-take-review/SKILL.md

video-take-review

Skill7 starsChanged 22 days agoMITView on GitHub

The per-seed READ of generated takes before the operator sees them: continuity across the take's own cuts and against the established shots FIRST, then the acceptance matrix (script beat, technical, anatomy, faces, product, eyelines, audio, named states at 4× on every frame, dignity), then voids and usable windows, then clips to the operator with residual doubts and frame times. Use when seeds land, a take must be judged against the ledger, the operator names a defect on a frame, a keeper's window must be set, or an instrument's number is about to be trusted. Triggers — "review the seeds", "which take", "continuity sheet", "is this seed usable", "cut before it breaks", "the same guy?", "frame 0:41", "check every frame", "break down the client's clips", "which window of this clip". Not for building the references the next shot needs — use video-refs-continuity. Not for the cost line or the submit — use video-gen-cost-gate. Not for placing the pick or applying a cut note — use video-edit-edl. Not for the delivered file's loudness or encode — use video-finish-qc.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Done when: every seed has a sheet, a cut list, a transcript and a peak time in review/, and the previous keeper's last frame (or the plate) …

    line 64
  2. Distinctive:

    Done when: every seed is VOID (reason) or KEEPER (window, in/out, the state at the window's end), and a void on a shot that has tripped a …

    line 161
  3. Distinctive:

    No numeric tolerance is invented.

    line 202

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Video Take Review

A landed seed is read the same day, by the agent, with instruments — and the operator's verdict is the only pick. The read leads with continuity — the priority consideration for accepting or rejecting a gen — never with the gag; a void carries its reason into the ledger; a keeper carries its usable window; a residual doubt travels in the delivery note with its frame time, not in a re-roll question. Four rules above the matrix: report what you SAW, not the win you hoped for (a take that fixes one defect and introduces a worse one has the worse one as its headline); a whole-frame downscale assesses composition only — identity, hands, teeth and product are read at 1:1 or larger; every instrument prints a known-answer self-test beside its number; and evidence is not looking — a take that probes valid and passes every measured row has still not been SEEN until its frames were read, so until then the report says "container valid · measured rows pass · visual pending", and appearance is never inferred from a filename, a prompt, a job history or ffprobe (pattern: piorunkulaga174/dsh-comfyui references/visual-verification.md).

What varies. The pixel numbers here — the face floor, the instruments' blind spots, the composed-cut multiples — were measured on 480p takes: count in the take's native pixels, and re-measure on a new generator's first takes. The product rows exist only when a product is on screen, and the brand's standards and the genre's tone set the dignity row. Continuity first, the matrix and the self-tests do not move — video-production/references/WHAT-VARIES.md.

1. Build the read for every seed

python3 scripts/qc_seed.py --out review --plate <accepted plate> takes/S02-G4-s1.mp4
bash    scripts/contact_sheet.sh takes/S02-G4-s1.mp4 4 4         # --at s,… · --beats <beat_grid / timeline JSON> · --phone (360 px, one a second)
python3 scripts/speech_timing.py takes/S02-G4-s1.mp4 --wpm 230,250   # a take that SPEAKS; see the delivery row
python3 scripts/seed_montage.py --out review/S02-G4-montage.jpg takes/S02-G4-s*.mp4   # every seed of the shot on ONE sheet
python3 scripts/attach_proof.py --root <project> --take S02-G4-s1   # every gated reference was SENT — the attachment row

The seeds of one shot are then read TOGETHER (seed_montage.py): the seeds as columns at the same fractions of their length, and a similarity table naming the near-duplicates — read one at a time, a near-duplicate looks like a new angle. A void's row names the lever the next generation pulls (ACCEPTANCE-MATRIX.md § The lever a miss pulls).

qc_seed.py gives the cut list (per-frame diff), the continuity sheet — the plate | the first frame of every cut | the last frame — a 1 fps tile, the native-audio RMS peak (the hit's time) and a whisper transcript. Read the sheet at 2–4×: every background element in every cut (one object checked is not all of them). The transcript is of the WHOLE take, windowed afterwards to the excerpt in question — a short excerpt transcribed alone loses the agreement between models that the full take keeps (spot-audio-assembly § 5). Client footage is a take too: the same sheet, cut list and transcript per clip, and its display shape from video-production/scripts/probe_sources.py before any crop. A whole PACK of supplied clips is broken down in one pass — scripts/footage_intake.py --src <dir> --out review/<pack>-intake writes, per clip, the record (hash, display raster, the delivery crop, cut candidates, motion with the conform's cadence read, colour steps), per-frame.csv, and three tiers of sheet (overview → which clips are one camera ROUTE · survey → what is in frame across the whole clip · every frame → a window's in and out), plus the stub for the written read. The procedure around it — routes, slots, windows, the pick — is ad-spot-preprod/references/FOOTAGE-CURATION.md.

Done when: every seed has a sheet, a cut list, a transcript and a peak time in review/, and the previous keeper's last frame (or the plate) sits beside them.

Write the record, read the record. qc_seed.py --record leaves <take>.review.json beside the take — the take's sha256 and byte size, every measurement with the instrument and rule that produced it, the sheet path, and notes[] for what the review decided (qc_seed.py <take> --note "f69: the lamp exits right, correct"). A resumed session reads the record first (--read: a size check, then the hash) and re-views strips only for the claim it is about to act on — a compaction re-derived one lamp read from image strips twice at image cost before this existed. A record whose bytes or hash no longer match the take is STALE: --read refuses it, and it is re-measured, never hand-edited fresh.

Write each image for what it is FOR. A SURVEY image — the continuity sheet, the tile, a contact or cut sheet — is JPEG q85 and may be downscaled; composition survives it. A VERDICT image — a face, a hand, a garment, a named state — is a crop of the thing judged at native size or larger, JPEG q95 4:4:4, never downscaled. A GENERATOR INPUT — a start image, a plate, a reference — stays PNG. An agent session can die on request BYTES long before its context fills (lossless full frames, ~350 KB each, replayed every turn), so crop to what is judged; never cap a verdict image's dimensions. references/INSTRUMENTS.md § Review images.

2. Continuity first — a break rejects the seed before its gag is judged

Against the previous keeper's last state and the established shots, per cut: the door (leaf, swing, distance, which side of the lens), the seats and who is nearest the camera, counts on shelves, what is in whose hands, wardrobe (accessories, layers), set dressing (the same objects), the cast (each named character once, counted at 2× across the crowd; nobody invented), the pose state (standing stays standing; fired stays fired; blanketed stays blanketed, no second bang), the product's shape at every scale it appears and its size beside a known object, named states (teeth out, a garment off, glass in the pane). A geometry break is not hideable; a missing beat is cheap to add. The axes: video-refs-continuity/references/CONTINUITY-AXES.md.

A cut the model composed inside ONE generation is a capability, never a defect. Consistency is free inside one generation and a gamble between two — it is what a long generation is bought for — so the read across a composed cut is whether the world survived it: the same faces, wardrobe, room, light and performance register. Never fail a take for containing a cut, and never grade the cut's motivation as a pass or a fail. A seam between two generations (an editor's join, a splice hidden inside a shot) is a different object, judged at the edit.

python3 scripts/cut_consistency.py takes/S02-G4-s1.mp4 --cuts 5.5,9.25 --out review --box "5.5:150,200,330,420:150,200,330,420"

cut_consistency.py measures light, contrast, white balance and palette across every cut as multiples of the take's own within-shot baseline, and writes the frames either side (a survey sheet) and the --box crops (verdict images) for the identity and wardrobe read — a measurement beside the eye, never a verdict.

Done when: each seed carries a continuity verdict per cut with the failing axis named, or "holds".

3. The acceptance matrix — a table, never a score

Rows per seed, each answered PASS / FAIL / note; a FAIL is never overridden by a good-looking frame (references/ACCEPTANCE-MATRIX.md):

rowthe read
script beatthe scripted action IS the action (the verb the script wrote, not the one the model chose); the cause is on screen before the reaction; nothing invented
technicaldims, fps, CFR, duration; the take's internal cuts listed for the edit — DESCRIPTIVE, never a fail
consistency across cutsper cut the model composed: the same faces, wardrobe, room, light and performance register either side — read on cut_consistency.py's sheet and crops beside its multiples; a break on any axis FAILs, the cut's presence never does
anatomyper person, in a per-person crop at 3×: 2 arms, 2 legs, 1 head, each hand on an arm, one prop per hand
facesundistorted, ≥ ~60 px of face height in the take's native pixels (measured at 480p) for any beat the faces carry — a FLOOR, not a target: a face over it was rejected as under-rendered, and face detail compares only at equal resampled size; identity against the character's close shots (a reconstructed face is plausible, not faithful)
attachmentevery reference the take was gated on was SENT to the model — attach_proof.py compares the gate's submission record with the receipt's request record (MISSING = FAIL: the take was made without it, however right it looks; UNPROVEN = a receipt from before 2026-09-29 or a fal take — read the venue's own job record) — and the product is VISIBLE in the take and matches its reference, read on the product row. A path or a name in a prompt is not proof the model read the image
productshape at 2× (the exact silhouette, haphazard angles, never aligned); proportion by a 4× crop beside the product photo (the pixel instrument is blind at 480p)
compositionnobody looks into the lens; the line-deliverer faces the camera, never back-to-camera; each gesture motivated in ITS framing; the reaction faces its cause with the matching emotion; no focus pull off the face; the camera holds its lock (NCC scale-match of the upper frame between frames)
audiowhisper: word-like events on an unscripted mouth = "shouting" (void); the scripted words intelligible (a near-homophone is a void); a generated tone at the head (spectral flatness < 0.3 with a stable peak); the hit's peak time
deliveryon a take that SPEAKS, the half the audio row does not cover — whether the performance is usable, which is the half an agent cannot hear. speech_timing.py reads four numbers off the word times: the SPAN against the duration (over ~0.3 s of tail silence = the model was given more seconds than the script fills and RUSHED; head silence is the same defect, and only that one is cheap to trim); every GAP at or over 0.25 s with its time (zero or one at a breath is good — a 0.7 s hold mid-line traces to a full stop in the prompt, which the model reads as permission to stop); the PACE as words over the SPAN, never over the duration, against the band the GENRE asks for (230–250 wpm is the creator/UGC band, and a brand read sits far below it — there is no universal band); and STUTTER CANDIDATES, an adjacent repeat or a bigram repeated inside a short window ("job infinitely and make infinitely"), which generated speech produces with no audio artefact at all, so they pass an intelligibility check and only a transcript read catches them. Candidates, never a verdict: the script says whether a repetition was written. A throwaway tail the prompt added after a full stop, to size a line to a venue's fixed duration step (video-prompt-dialects PHONE-NATIVE.md § Dialogue), is not the performance: the window ends at the pause after the real line, and the tail is never read as a stutter or a rush
named stateevery frame of every window after the event, wides included, at 4× with the gamma lifted in dark cavities — a sparse sample passes defects that every frame at 4× shows
dignityawake; nobody touched unless the script does it; nobody left lying under an effect; hands high, visible, held; no hand at the bottom edge of a chest-up two-shot — the brand's standards and the genre's tone set the bar
realismposes natural; dance not jerky; no phantom extra; the runner seen from behind; the subject still in the last frame; on a creator-style take the tells pass — sound off first, hands on the product, a torso static > 6–7 s, consonant lip drift, the product morphing between shots, background warp, a phone or camera UI in frame (ACCEPTANCE-MATRIX.md row 10)

Done when: the matrix is filled for every seed, the worst row is the headline of each seed's line, and no row was inferred from a downscale.

4. Voids and windows

A seed that fails a hard row is VOID with its reason in the ledger (never reused). The same realism failure on several seeds of ONE reference — waxy skin, a dead mouth, a face that reads fake — is a reference void: the reference is rebuilt and re-accepted before the next seed (video-refs-continuity REPAIR-LADDER.md § Replace the reference, not the seed), because re-rolling against it spends seeds on a picture the model cannot animate. A keeper is a take plus a window: in/out set from the read — "only from 1/3 in, as they snap to", "cut before the teeth show", "only the first 0.9 s", "cut on the frame the set leaves the lips", the out-point at least one frame before the take's own internal cut. A window whose last state hides the face cannot be continued by a gen; say so now.

python3 scripts/window_frames.py --in takes/S01B-CCU2-s2.mp4 --from 2.9 --to 3.3 --zoom 4 --gamma 1.6 --box 120,300,360,520 --out review/ccu2-mouth.png
python3 scripts/frame_match.py --take takes/S01B-F4-s1.mp4 review/operator-frame.png      # "which event is this?"

A window of supplied footage is ranked at the DELIVERY shape, never on the source frame. scripts/window_metrics.py --root --candidates <json> --out --intake <intake dir> --stack decodes each candidate once through its delivery crop and raster and prints, per metric with its validity: sharpness (comparable only inside one camera route), luma and clipping, colourfulness, speed in frame-widths per second, the frame-to-frame step in pixels at the delivery width — on a dropped-frame conform the recurring double step is twice it: under ~5 px it does not read, near 30 px a pan stutters — and shake on the axis the move does not use. The 5-frame strip is the verdict; a window that opens on the wrong thing fails on its first frame.

A void names whether the next take is a re-roll at all. The same flaw on two takes of an unchanged prompt stops the re-rolls: the next take changes ONE variable, the lever ACCEPTANCE-MATRIX names for that row. A shot at three paid attempts with no pick and no declared budget goes to the operator as options, not as another batch (video-gen-cost-gate COST-AND-GO.md § Budget-final rules, the stop-rule ladder; take_ledger.py reads the counts).

Done when: every seed is VOID (reason) or KEEPER (window, in/out, the state at the window's end), and a void on a shot that has tripped a stop rule says so.

5. Clips to the operator; the pick is theirs

Send the clips by path the moment they land (chat send only under the size cap, else the Windows path); the sheets are the agent's notes, one line per seed: headline row, window, residual doubts with frame times ("a wedding band 2.4–3.3 s"). The operator picks by path. When the operator names a defect: locate it on THEIR frame (frame_match.py against the take windows and the gen heads), confirm the person and the event back, and only then act — a defect acted on before it was located costs a wrong fix or a wasted regen. When the operator says they see it in motion, they see it.

Done when: the pick is recorded by path with its window in the continuity ledger and in code (pick_gate.py record --root <project> --window <in>-<out> <path>, video-production — the upscale and hero scripts refuse a clip that traces to none), voids carry reasons, and the next skill (the ledger update in video-refs-continuity, then the finishing chain after approval) can start from the record alone.

❌/✅

❌ "s2 plays the whole beat" as the headline                 ✅ "s2 breaks the door wall; s1 holds — s1"
❌ A contact sheet as the review                               ✅ The clips by path; the sheet as notes
❌ Identity judged on a 1000 px downscale                      ✅ A 1:1 crop of the face
❌ One figurine zoomed in one cut → "background clean"        ✅ Every element in every cut
❌ A 1.5× sample of the mouth window                           ✅ Every frame at 4×, gamma lifted
❌ "fixed the double arm" on a different person                ✅ Locate on the operator's frame, confirm back
❌ "Approve, or re-roll for the fleck?"                        ✅ The fleck in the delivery note with its time
❌ An instrument's number with no self-test                    ✅ A known-answer case printed beside it
❌ "the wide has no readable teeth"                            ✅ The named state read in the wide too, at hero size
❌ A take VOIDed because it contains a cut                     ✅ The cut listed; faces, wardrobe, room, light, register read across it
❌ A lossless full frame as a review image                     ✅ A crop of what is judged at q95; surveys at q85
❌ "the face clears 60 px" as a pass                           ✅ The floor cleared; detail compared at equal size
❌ "jitter 2.6 px - stabilise the pan"                         ✅ The residual ALONG a pan reads the conform cadence; shake is on the other axis
❌ 17 supplied files ranked as 17 shots                        ✅ The route table first: one subject + one move = one shot, whatever the file count

Failure behavior

  • An instrument that cannot detect the defect (a scale metric that matched a frame to itself at 0.90×; a hair-blob metric contaminated by loose pieces) is reported as broken, not as a number.
  • No numeric tolerance is invented; the assets and the script define the target.
  • Three non-converging rounds on one element → stop generating; the operator gets the alternatives as numbered questions.
  • Nothing is upscaled, hero-passed or cut on a seed before the operator's pick.

Cross-references

  • video-refs-continuity — the axes, the ledger, the repair ladder when a read fails.
  • video-gen-cost-gate — budget-final rules (a seed meeting every named constraint goes forward).
  • references/INSTRUMENTS.md — every script, what it measures, its self-test.
  • Deeper context — the upstream projects behind the rules here: video-production/references/CONTEXT-MAP.md § Where the deeper context lives.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.