SkillAgentSearch skills...

video

Video projects - list, resume, or create new

Install / Use

npx skills add digitalsamba/claude-code-video-toolkit

Installs into whichever agent you are using.

About this skill
⚡

Claude Commands

Claude Code slash commands

Quality Score

88/100

Category

Other

Supported Platforms

Claude Code

Our assessment of video

video scores 88/100 on our quality scale, 134th of 331 Other skills we index (top 41%).

Its Claude Commands is 17 KB long, well organised into 41 sections with 19 code examples: a thorough specification that gives an agent plenty to work with.

With 2,544 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
30/30
Structure
20/20
Description
8/15
Adoption
14/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 5 days ago, so video is actively maintained.
  • Our last check on 2026-08-24 found the source still online.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

video compared with similar skills

All 4 of these similar skills score higher than video; compare them before choosing.

SkillScoreStarsUpdatedFormat
video (this skill)by digitalsamba882.5k5d agoClaude Commands
claude-memby thedotmack100117.9ktodayCLAUDE.md
taste-skillby Leonxlnx100117.6k9d agoCLAUDE.md
Agent-Reachby Panniantong100112.7k9d agoCLAUDE.md
Understand-Anythingby Egonex-AI10093.0k2d agoCLAUDE.md

Frequently asked questions

How do I install video?
Run npx skills add digitalsamba/claude-code-video-toolkit. The install tabs above show the steps for each supported agent.
Which AI agents does video work with?
It is written for Claude Code, as a Claude Commands file. Other agents that read the same format can often use it too.
Is video safe to use?
It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is video still maintained?
The repository was last updated 5 days ago, so video is actively maintained.

description: Video projects - list, resume, or create new

Video

Unified command for video projects. Scans for existing projects and offers to resume or create new.

Entry Point Logic

On invocation, scan for projects and adapt:

Step 0: Check Setup (first-time only)

If .env does not exist or no cloud GPU provider is configured (no MODAL_*_ENDPOINT_URL or RUNPOD_*_ENDPOINT_ID values set):

Tip: Run /setup to configure AI voiceovers, image generation, and music.
It takes ~5 minutes and most features are free. You can skip this and
create videos with just Node.js — run /setup anytime later.

This is a one-line hint, not a blocker. Proceed to Step 1 immediately.

Step 1: Scan Projects

1. Glob projects/*/project.json
2. For each project found:
   - Read project.json
   - Check filesystem for actual assets (public/demos/*, public/audio/*)
   - Reconcile status vs reality
   - Calculate health (ready/blocked/stale/complete)
3. Sort by last updated (most recent first)

Step 2: Present Options

No projects found:

No video projects found. Let's create one!

Which template would you like to use?
  1. Sprint Review - internal demos, release videos
  2. Product Demo - marketing, launches, features
  3. Concept Explainer Short - vertical TikTok/Reels/Shorts explainers
  4. DS Release Notes - Digital Samba release videos, screens rebuilt from source

→ Proceed to New Project Flow

One project found:

Found: **my-release-video** (sprint-review)
  Phase: assets
  Progress: 2 of 5 demos recorded
  Last worked: 2 days ago

  → Resume this project
  → Start a new project

Multiple projects found:

Found 3 video projects:

  1. **my-release-video** (sprint-review)
     Phase: assets - 2/5 demos recorded
     Last worked: 2 days ago

  2. **product-launch** (product-demo)
     Phase: audio - voiceover needed
     Last worked: 5 days ago

  3. **tutorial-v1** (sprint-review)
     ✅ Complete
     Rendered: 2 weeks ago

Which project? (or 'new' for a new project)

Resume Flow

When resuming a project:

Step 1: Read State

1. Read project.json
2. Read VOICEOVER-SCRIPT.md for context
3. Glob public/demos/* and public/audio/*
4. Compare to project.json expectations

Step 2: Reconcile Filesystem

For each scene with visual.asset:

  • If status: "asset-needed" but file exists → update to "asset-present"
  • If status: "ready" but file missing → update to "asset-missing", flag issue

For audio:

  • Check if public/audio/voiceover.mp3 exists
  • Update audio.voiceover.status accordingly

Step 3: Present Current State

Resuming: my-release-video (sprint-review)

## Scenes

| # | Scene | Type | Status |
|---|-------|------|--------|
| 1 | Title | title | ✅ Ready |
| 2 | Overview | overview | ✅ Ready |
| 3 | Dark Mode Demo | demo | ✅ Recorded |
| 4 | Login Flow | demo | ⬜ Needs recording |
| 5 | Stats | stats | ✅ Ready (slide) |
| 6 | Credits | credits | ✅ Ready (slide) |

## Audio

- Voiceover: ⬜ Not yet generated (script ready)
- Music: Optional

## Next Actions

1. **Record login flow demo** (Scene 4)
   Run `/record-demo` or provide external video

2. **Generate voiceover** (after demos complete)
   Run `/generate-voiceover`

(If the project has no `design.md`, add "Set the design direction: run `/design`" before building more scenes.)

Ready to record the login flow?

Step 4: Update project.json

After reconciliation:

  • Update any changed statuses
  • Add session entry: { date: "YYYY-MM-DD", summary: "Resumed project" }
  • Update updated timestamp

Step 5: Regenerate CLAUDE.md

Update the project's CLAUDE.md with current state (see CLAUDE.md Generation below).


New Project Flow

When creating a new project:

Phase 1: Setup

Template Selection:

Which template would you like to use?

| Template | Best For | Scene Types |
|----------|----------|-------------|
| Sprint Review | Internal demos, release videos | title, overview, demo, split-demo, summary, credits |
| Product Demo | Marketing, launches, features | title, problem, solution, demo, feature, stats, cta |
| Concept Explainer Short | Vertical 9:16 TikTok/Reels/Shorts explainers | hook, concept cards (Ideogram), b-roll (LTX), recap, cta |
| DS Release Notes | Digital Samba release videos, nothing screen-recorded | title (item list), one scene per release item, outro |

DS Release Notes carries its own Digital Samba look (room and account-center kits), so skip the brand question and the brand.ts and brand-asset steps below. Name the project release-<number> and set RELEASE_NUMBER and RELEASE_MONTH in src/config.ts. Scenes come from the release notes, not the scene planner: follow "Making the next release" in templates/ds-release-notes/README.md and the synthetic-ui-demo skill.

Concept Explainer Short is Python-based (moviepy, no Remotion/npm). Its flow differs after scene planning: author scenes.json → generate visuals (ideogram4/ltx2 from toolkit root) → gen_vo.py → gen_captions.py → build.py. See templates/concept-explainer-short/README.md. Skip the npm-install and brand.ts steps below; preview = open out/short.mp4.

Brand Selection:

List available brands from brands/ directory:

1. Read each brands/*/brand.json
2. Extract name and primary color
3. Present list with "Create new brand" option
Which brand?

  1. default - #3B82F6 (blue)
  2. digital-samba - #FF6B00 (orange)
  3. Create new brand → /brand

Project Name:

What should we call this project?
(This becomes the folder name, e.g., "v21-release" → projects/v21-release/)

Phase 2: Content Gathering

What content should this video cover? You can:
- Paste release notes or bullet points
- Provide a URL (release notes, changelog, JIRA, GitHub)
- Describe what you want to show
- Any combination of the above

If URL provided:

  • WebFetch the content
  • Parse and summarize
  • Ask clarifying questions

Always ask:

  • "What's the main message or announcement?"
  • "Who's the audience?" (internal team, customers, developers)
  • "Any specific features to highlight?"
  • "Anything to skip or de-emphasize?"

Phase 3: Scene Planning

Before proposing scenes, see whether Blender is here (optional; most installs won't have it):

uv run tools/blender_render.py doctor --no-bench   # exit 0: Blender 4.2+ and ffmpeg are ready
  • Exit 0: where a scene fits one, propose a Blender shot beside the usual visuals, and say why: a character reacting to the news (a gag, a before and after), the scene turning out to be on a laptop as the camera pulls back, the app on a phone or a floating card, a 3D model of the product. One or two in a video is plenty. A video that is 3D from start to finish needs no template: it is one shot spec in the project folder.
  • Not found: say nothing and plan as usual. If the user asks for 3D, tell them it needs Blender (docs/blender.md, a free install) and offer the nearest thing without it (an LTX clip, synthetic-ui-demo).

The spec format and shot grammar are in the blender skill.

Step 1: Propose scene breakdown

Based on content, propose scenes:

Based on your input, here's a suggested scene breakdown:

  1. Title (5s) - "Version 2.1 Release"
     [SLIDE: Title with logo]

  2. Overview (15s) - Key highlights
     [SLIDE: 3 bullet points]

  3. Demo: Dark Mode (20s)
     [DEMO: Settings → toggle → show UI change]
     "One of our most requested features..."

  4. Stats (10s)
     [SLIDE: 3 features, 5 fixes]

  5. Credits (5s)
     [SLIDE: Team credits]

  Estimated total: ~55 seconds

Want to adjust any scenes?

Step 2: Iterate

Help refine:

  • Add/remove scenes
  • Reorder
  • Adjust timing
  • Change visual type

Step 3: Confirm each scene

| Aspect | Options | |--------|---------| | Type | title, overview, demo, split-demo, stats, credits, problem, solution, feature, cta | | Visual | slide, playwright, external, screenshot, blender (when installed: a shot spec, rendered as a clip) | | Narration | Draft voiceover text | | Duration | Estimated seconds |

Phase 4: Design Direction

Decide how the video looks before any scene is built, so the scenes are made from the design instead of from the template's defaults and restyled afterwards. The scene plan is enough to work from: what each scene says, where the story turns, and whether it's 16:9 or 9:16.

  1. Design source. Ask as in /design "Design source" (ours, the installed registry designSources, another design skill). Skip the question when the brand or template already fixes the look (DS Release Notes carries its own kits): record "ours" and go on.

  2. Direction for the whole video. Invoke the source with the scene plan, the audience, the format and the brand (for ours, pick an act palette from .claude/skills/video-design/palettes.md). Then invoke video-design to map it onto the plan: group the scenes into acts at the story's turns, a field per act, ink, the one accent, the five type steps, a ground from the subject if a field would otherwise be dull.

  3. Show it before writing it. A short summary, one line per act, so the user can redirect cheaply:

    Design: taste-skill, direction "colour chapters"
    Act 1 (scenes 1-2, the problem)   near-black field, bone ink, title 120px
    Act 2 (scenes 3-5, the feature)   brand field, black ink, the product window on a hairline edge
    Act 3 (scene 6, call to action)   bone field, black ink, the URL at the figure size
    Accent: brand colour, on the one number the narration lands on
    

The answers are written in Phase 5: design.md (format in .claude/skills/video-design/handoff.md) and project.json design. /design can redo this step later, or run it for a project made without it.

Phase 5: Project Creation

  1. Copy template:

    cp -r templates/{template}/ projects/{name}/
    
  2. Generate project.json: Use schema from lib/project/types.ts:

    {
      "name": "{name}",
      "template": "{template}",
      "brand": "{brand}",
      "created": "{ISO timestamp}",
      "updated": "{ISO timestamp}",
      "phase": "planning",
      "design": { "source": "{registry key, or the skill's name}", "file": "design.md" },
      "scenes": [...],
      "audio": {
        "voiceover": { "file": "audio/voiceover.mp3", "status": "needed" }
      },
      "estimates": { "totalDurationSeconds": N },
      "sessions": [{ "date": "YYYY-MM-DD", "summary": "Project created" }]
    }
    
  3. Generate brand.ts: Read brands/{brand}/brand.json → generate src/config/brand.ts

  4. Write VOICEOVER-SCRIPT.md: Full script with narration and asset markers

    Write design.md from Phase 4, and build the slide scenes from it: its fields, ink, accent and type steps replace the template's theme values where they differ (put them in the theme or brand config, not per scene).

  5. Copy brand assets: Logo from brands/{brand}/assets/ to public/images/

  6. Generate CLAUDE.md: Auto-generated project status (see below)

  7. Install dependencies:

    cd projects/{name} && npm install
    

    (Skip for concept-explainer-short — Python template, no npm.)

Phase 6: Asset Guidance

After creation, actively guide asset creation:

Project created at: projects/v21-release/

You have 2 assets to create:

  ⬜ Scene 3: Dark mode demo (20s)
     → Record with /record-demo or provide external video

  ⬜ Scene 4: Stats dashboard screenshot
     → Capture with Playwright or provide image

  ⬜ Scene 5: The app on a phone, camera orbiting (Blender)
     → Write blender/phone.json, `blender_render.py check`, read the contact sheet, then
       `render --clips --alpha --to projects/<name>/public/blender`

Which would you like to start with?

Track progress:

  • Update project.json as assets co

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars2.5k
CategoryOther
Updated5d ago
Forks444

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions