konradre/video-production-skills / skills/client-rounds
client-rounds
VerifiedRuns a CLIENT ROUND on delivered generated-video work: the notes saved verbatim and mapped to spot ids, every cited beat located in the delivered file, the EDL history and the script before anything moves, the notes curated into a numbered list (CUT, RECYCLE, REBUILD, CAUSALITY, APPROVE, GATE; each an ERROR or TASTE, anchored, and every uncaught ERROR given its check) the operator answers per item, the fixes executed by the ladder into a NEW version with approved spots frozen, a compare page of the new version against the last lined up by event, and the ask — every note answered by its number, the checks read from the run log — as numbered questions with cost and path inline. Use when client feedback arrives, a delivered spot must be revised, a deliverable is about to be sent, or the operator must choose. Triggers — "client feedback", "the client's notes", "curate the list", "what did the client mean", "approved as is", "alternate ending", "send this to the client", "which option", "lock these in", "finals for the client". Not for the cut itself — use video-edit-edl. Not for the render or the QC instruments — use video-finish-qc. Not for a new brief or a new spot — use ad-spot-preprod.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 14.8 KB |
| references/ROUND-LEDGER.md | markdown | 1.3 KB |
| references/ROUND-PROTOCOL.md | markdown | 10.6 KB |
| references/THE-ASK.md | markdown | 2.6 KB |
| scripts/compare_versions.py | script | 28.5 KB |
| scripts/delivery_ask.pyexecutable | script | 18.2 KB |
| scripts/notes_triage.pyexecutable | script | 28.9 KB |
references/THE-ASK.md
View on GitHub# The ask — what the operator reads, and what the client receives
## To the operator (decisions)
- **Numbered plain questions**, each answerable with a number or a yes/no, **the cost inline**
("1. Redo scene C by extending scene B (3 seeds, 37.5 cr) — yes or no? 2. Which clip, 2–7? 3. Add
the door-slam cutaway (3 seeds, 37.5 cr) — yes or no?"); a default when there is one ("say
defaults"); rationale before or after, never inside the questions.
- **Every item to look at carries its full path IN THE ASKING MESSAGE** (as the operator opens it); a pick-a-number
question lists one line per option = label + path (+ the tile or reel path); re-asking after a pause
repeats the paths.
- Seeds go to the operator as **clips**, never contact sheets; "not bad" is not a pick — wait for the
path.
- A file ≤ 30 MiB may ride in chat; above that, the path is the delivery (uploads over ~20 MiB time
out). 1080p masters only.
## With the deliverable (`delivery_ask.py`)
1. The spot and version, the full path, runtime and size.
2. What changed since the previous version (one line per item, mapped to the client's numbered notes) — in a round, every
note answered by its number from the round record (`--round`), and the checks proposed for its errors.
3. The VO script as placed — every line verbatim with its time.
4. The checks, read from the run log, never retyped (`--log`; `--qc` counts as one): the last `EDL-CHECK` and `BEAT-SHEET`
line printed for this EDL and the last `QC-DELIVERABLE` line for this file. A check with no line refuses the ask;
`--na "<check>: <why>"` covers one that does not apply. A FAIL is shown, never withheld — QC's judgement rows are the
operator's call. The beat sheet (`beat_sheet.py`) is pasted as well, so the cut is read against the script without
opening the file.
5. The compare page's path when the version revises an earlier one (`compare_versions.py`; `--compare`): N against N−1,
lined up by event.
6. **Residual doubts with frame times, in the note** — never "approve or re-roll over a fleck?" after a
final top-up; the operator decides whether the client sees them.
7. What is frozen (approved spots, in the client's words).
8. The decisions, numbered.
## Saving is not sending
Every round's deliverables land in the standing deliver directory (a named output directory is a
standing instruction for the whole arc, never re-asked) with their provenance beside them (the EDL,
the notes file, the receipts); a chat attachment is not a save. Sub-folders per round keep it legible
(`deliver/`, `deliver/final/`, `deliver/alts/`).
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |