video
Video projects - list, resume, or create new
Install / Use
npx skills add digitalsamba/claude-code-video-toolkitInstalls into whichever agent you are using.
Claude Commands
Claude Code slash commands
Quality Score
Category
OtherSupported Platforms
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.
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.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| video (this skill)by digitalsamba | 88 | 2.5k | 5d ago | Claude Commands |
| claude-memby thedotmack | 100 | 117.9k | today | CLAUDE.md |
| taste-skillby Leonxlnx | 100 | 117.6k | 9d ago | CLAUDE.md |
| Agent-Reachby Panniantong | 100 | 112.7k | 9d ago | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 93.0k | 2d ago | CLAUDE.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.
Skill content
View source on GitHubdescription: 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.mp3exists - Update
audio.voiceover.statusaccordingly
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
updatedtimestamp
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 setRELEASE_NUMBERandRELEASE_MONTHinsrc/config.ts. Scenes come from the release notes, not the scene planner: follow "Making the next release" intemplates/ds-release-notes/README.mdand thesynthetic-ui-demoskill.
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. Seetemplates/concept-explainer-short/README.md. Skip the npm-install and brand.ts steps below; preview = openout/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.
-
Design source. Ask as in
/design"Design source" (ours, the installed registrydesignSources, 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. -
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 invokevideo-designto 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. -
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
-
Copy template:
cp -r templates/{template}/ projects/{name}/ -
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" }] } -
Generate brand.ts: Read
brands/{brand}/brand.json→ generatesrc/config/brand.ts -
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).
-
Copy brand assets: Logo from
brands/{brand}/assets/topublic/images/ -
Generate CLAUDE.md: Auto-generated project status (see below)
-
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
claude-mem
117.9kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
taste-skill
117.6kAgent skills that give coding agents taste: art direction, reference images and a browser check instead of generic AI pages
Agent-Reach
112.7kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
Understand-Anything
93.0kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
Languages
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
