Skip to content
Agentmart
Submit a tool

video-workspace-conventions

Reference for the AI video production workspace — lifecycle phases, folder layout, naming conventions, and working principles.

Skill by shotcowboystyle

IndexedMIT
Write a review

Chalk Labs

Need help deploying video-workspace-conventions?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/shotcowboystyle-video-workspace-conventions

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Reference for the AI video production workspace — lifecycle phases, folder layout, naming conventions, and working principles. Loaded by the install-video-workspace command and consulted by other commands when they need to know where things live. Use when you need to know where a file belongs or which lifecycle phase applies.

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/video-workspace-conventions

What's inside

Skill name
video-workspace-conventions
Description
Reference for the AI video production workspace — lifecycle phases, folder layout, naming conventions, and working principles. Loaded by the install-video-workspace command and consulted by other commands when they need to know where things live. Use when you need to know where a file belongs or which lifecycle phase applies.
License
MIT
Instructions
78 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# AI Video Production Workspace Conventions

A workspace produced by `/install-video-workspace` follows the conventions below. All other commands in this plugin assume this structure.

## Lifecycle and where things live

| Phase | Folder | Contents |
|-------|--------|----------|
| Brief | `brief/` | Creative brief, target spec, model/tool selection, vibe references |
| Characters | `characters/` | One markdown file per recurring character/subject — appearance, voice, reference image links |
| Script | `scripts/` | `drafts/` (working), `final/` (locked), `storyboards/` (shot-by-shot breakdowns), `tts/<model>/` (TTS-ready VO chunks) |
| Source assets | `assets/` | Reference images, voice samples, voice-clone source audio, SFX, music, fonts, LUTs, overlays |
| Generation output | `generation/` | Raw model output by modality: `text-to-image/`, `image-to-image/`, `image-to-video/`, `text-to-video/`, `voice/`, `lip-sync/`, `upscale/`. Prompts that produced each artefact go in `generation/prompts/` |
| Pipelines | `pipelines/` | One directory per pipeline; see [Pipeline layout](#pipeline-layout) |
| Clips | `clips/` | `raw/`, `selected/`, `edited/`, `b-roll/`, `normalised/` (loudness/framerate-conformed copies), `converted/` (aspect/resolution-converted copies) |
| Timeline | `timeline/` | Shot list, edit decision list, sequencing notes |
| Exports | `exports/` | Thumbnails, captions, subtitle files (.srt/.vtt) |
| Output | `output/` | `drafts/` (rough cuts), `proofs/` (review copies), `final/` (locked deliverable) |
| Budgets | `budgets/` | Cost estimates from `budget-estimator` (`estimate-<YYYY-MM-DD>.md`) |
| Research | `research/` | Model shortlists and recommendations (`models-*.md`, `recommendations-*.md`) |
| Logs | `logs/` | Per-session production logs, model-call records, cost tracking |

## Pipeline layout

Every pipeline is a directory. Commands, agents and skills all read and write this layout:

```
pipelines/<name>/
  SPEC.md                  — the plan: stages, models, inputs/outputs, cost envelope, open questions
  run.md                   — numbered steps `/run-pipeline <name> <shot#>` follows
  README.md                — optional: how to invoke, expected runtime and cost
  stages/NN-<stage>.md     — optional: prompt template + parameters per stage
  parameters.example.json  — optional: template parameters; copy to parameters.json per shot
  runs/                    — per-run records
  comfy/                   — optional: ComfyUI workflow and manifest
```

- `ai-video-producer-pipeline-scaffolder` writes `SPEC.md`, `stages/` and `runs/`.
- `ai-video-producer-pipeline-builder` writes `run.md`, `README.md`, `stages/*.md` and `parameters.example.json`.
- `/run-pipeline` needs `run.md`. A pipeline with only `SPEC.md` is not built yet.

## Working principles

- **Brief before generation.** Don't fire model calls until `brief/creative-brief.md` exists with target runtime, format, aspect ratio, and model selection. If generation starts without a brief, prompt the user to run `/onboard`.
- **Pair every artefact with its prompt.** Whenever a generated image/clip/voice take is saved, also save the prompt + model + seed in `generation/prompts/` with a matching filename stem so a take can be reproduced or iterated.
- **Raw vs selected.** Everything the model produces lands in `generation/<modality>/` and `clips/raw/`. Only takes the user explicitly approves get copied into `clips/selected/`. Never delete raw output without asking — iteration loops back to it.
- **Pipelines are recipes, not magic.** A pipeline file describes a chain (e.g. Flux text-to-image → Kling image-to-video → ElevenLabs voice → lip-sync). When a pipeline is invoked, walk through it step-by-step, surfacing the intermediate outputs for approval.
- **Tools are user-chosen.** Model selection is per-project. Read `brief/tools-and-models.md` for the active selection — don't assume defaults.
- **Log model calls.** Append a one-line entry to `logs/production-log.md` for each non-trivial generation: timestamp, model, cost estimate, output path.

## Naming conventions

- Generated artefacts: `NN-shortname-vN.ext` — e.g. `03-rooftop-establish-v2.mp4`. The `NN` orders shots, `vN` tracks iterations.
- Prompts: matching stem in `generation/prompts/` — e.g. `03-rooftop-establish-v2.md` containing prompt + model + parameters.
- Characters: `characters/<name>.md`.
- Pipelines: `pipelines/<kebab-name>/` (see [Pipeline layout](#pipeline-layout)).

## Slash commands (typical order)

1. `/install-video-workspace` — scaffold this structure into a new directory.
2. `/onboard` — capture the creative brief and tools-and-models selection.
3. `/define-character` — add recurring characters/subjects.
4. `/draft-script` — produce a script draft from the brief.
5. `/storyboard` — break the locked script into a shot list.
6. `/run-pipeline <name> <shot#> [--dry-run]` — execute a pipeline for a shot.
7. `/log-take` — record a generated artefact.
8. `/promote-take` — promote a raw take into `clips/selected/`.
9. `/assemble` — propose an ffmpeg edit plan into `output/drafts/`.
10. `/export-final` — encode locked draft to spec, write to `output/final/`.

## What NOT to do

- Don't generate cover art, script, or clips before the brief exists.
- Don't overwrite files in `clips/selected/` or `output/final/` — version with `-v2` suffix instead.
- Don't invoke a model the user hasn't listed in `brief/tools-and-models.md` without asking.
- Don't delete from `generation/` or `clips/raw/` — discarded takes often get revisited.

From shotcowboystyle/ag-skills, MIT.

More from shotcowboystyle/ag-skills

Signals

Stars
0
Forks
0
Open issues
29
Last commit (this folder)
Oct 23, 2026 (8 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from shotcowboystyle

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get video-workspace-conventions running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying video-workspace-conventions?

CallGet help
Cite this page

video-workspace-conventions skill. Agent Mart. https://agentmart.dev/skills/shotcowboystyle-video-workspace-conventions (last updated 2026-10-24)