VidCraft Skills - Detail Reference
All 33 VidCraft skills by production phase: invocation with arguments, purpose, result and model used. Always invoked as /vidcraft:skill-name.
VidCraft Skills - Detail Reference
There are 33 skills. You invoke them as /vidcraft:<skill>. Arguments in square brackets are optional. <project> is the project slug, <episode> the episode slug.
The model is declared in each skill's frontmatter. Creative work and research run on Opus, as do storyboard-creator and video-reviewer. The remaining checks, formatting and project management run on Sonnet, quick lookups (help, timing-validator) on Haiku.
All skills except setup, configure, help and video-type-creator talk to the vidcraft-mcp MCP server, see Technical.
Setup and navigation
| Skill | Arguments | Purpose | Model |
|---|---|---|---|
setup |
- | First-time setup: Python environment, dependencies, configuration template | Sonnet |
configure |
- | Create or change config.yaml interactively, create directories |
Sonnet |
session-start |
- | Check plugin version, configuration and project list, rebuild the cache, show last session and pending ideas | Sonnet |
help |
- | Overview of the main skills and the typical flow | Haiku |
new-project |
<title> <video-type> |
Create a project (default: German, HeyGen), optionally episodes, remember it as the active project | Sonnet |
resume |
<project> |
Find the project, show progress, recommend the next step | Sonnet |
next-step |
[project] |
Exactly one recommended action based on status | Sonnet |
project-dashboard |
<project> |
Progress per episode, status breakdown, blocking items | Sonnet |
Concept and research
| Skill | Arguments | Purpose | Result |
|---|---|---|---|
doc-analyzer |
<file-path> [video-type] |
Analyze documentation, assess complexity, suggest structure. See Document analysis | Report in chat, optionally a project or idea |
researcher |
<topic-or-project> |
Web search on topic, competitor videos, technical accuracy | research/RESEARCH.md |
audience-researcher |
<project-or-topic> |
Personas, pain points and audience vocabulary via web search | research/AUDIENCE.md |
project-conceptualizer |
<project> |
Concept in one pass: batched questions, episode list, visual identity | Updated project README, created episodes, status Brief Complete |
brief-creator |
<project> |
Creative brief based on templates/brief.md |
brief.md, status Brief Complete |
video-type-creator |
<type-name> |
Research and document a new video type via web search | video-types/<type>/README.md in the plugin directory |
Script and review
| Skill | Arguments | Purpose | Model |
|---|---|---|---|
script-writer |
<project> <episode> |
Write scenes with narration, on-screen text and visual direction, check timing and readability, set status Script Draft |
Opus |
script-reviewer |
<project> <episode> |
14-point checklist plus pronunciation advisory, status Script Reviewed when everything passes |
Sonnet |
voice-checker |
<project> <episode> |
Scan narration for AI-sounding language (hints only, no rewriting) | Sonnet |
timing-validator |
<project> <episode> |
Quick check of total duration, scene lengths, hook and speaking pace | Haiku |
Visuals and assets
| Skill | Arguments | Purpose | Result |
|---|---|---|---|
storyboard-creator |
<project> <episode> |
Per scene: visual type, avatar position, background, transitions, timing | Storyboard file, enriched scenes |
screenshot-planner |
<project> <episode> |
Define required screenshots, screen recordings and graphics | Episode assets/ASSETS.md |
asset-collector |
<project> <episode> |
Check ASSETS.md against available files | Report, status Assets Collected when complete |
shot-list-creator |
<project> <episode> |
Break scenes into individual shots | Episode shot-list.md |
avatar-selector |
<project> |
Avatar and voice recommendation by audience, language and type | Recommendation, entry in the project README |
Production
| Skill | Arguments | Purpose |
|---|---|---|
pre-generation-check |
<project> <episode> |
All gates before generation, blocks on critical errors. See Quality system |
heygen-engineer |
<project> <episode> |
Format the script for HeyGen, detect variables, recommend Voice Director and motion prompts, list post-production tasks |
synthesia-engineer |
<project> <episode> |
Format the script for Synthesia slides, choose layouts, handle gesture tags |
Generation itself does not happen in VidCraft. Details: Platforms.
After generation and publishing
| Skill | Arguments | Purpose | Result |
|---|---|---|---|
video-reviewer |
<project> <episode> |
20-point review after generation | Report, status QA Passed when everything passes |
brand-checker |
<project> <episode> |
Tone, terminology, visual identity | Report |
accessibility-checker |
<project> <episode> |
10 points on subtitles, contrast, comprehensibility | Report |
subtitle-generator |
<project> <episode> |
Whisper transcription (German), correction, English translation | {episode}_de.srt and {episode}_en.srt in the assets folder |
thumbnail-director |
<project> <episode> |
Thumbnail concept with composition, text and image prompt | Concept in chat |
promo-writer |
<project> [platform ...] |
Promo copy for YouTube, Facebook, Instagram, LinkedIn, Twitter/X, email and wiki embed | promo/<platform>.md |
release-director |
<project> |
Release checklist, metadata per episode, channels, status Published |
RELEASE.md |
The order between skills is described on the Workflow page.
next-stepandresumeread the status-to-skill mapping fromknowledge/status-workflow.md.