ai-video-generation - Setup Guide¶

Source: 101-skills/skills (196,300 installs) Category: AI Media / Video Production Languages: TypeScript, Python

Production-grade AI video generation skill for coding agents. Complete script-to-video pipeline: generate scripts, compose scenes, integrate AI avatars with lip-sync, synthesize voiceovers, and export in multiple formats. Companion skill ai-avatar-video (195.8K installs) adds synthetic presenter capabilities. At 196K installs, this is the most-installed AI video skill in the ecosystem.

Installation¶

# Core video generation
npx skills add 101-skills/skills@ai-video-generation

# AI avatar presenters (optional, recommended)
npx skills add 101-skills/skills@ai-avatar-video

Verify:

npx skills list | grep "101-skills"

Prerequisites¶

Requirement	Details
Node.js	18+
FFmpeg	5.0+ (`sudo apt install ffmpeg` or `brew install ffmpeg`)
Hermes Agent	Any version
GPU (optional)	Accelerates rendering; CPU fallback available
HeyGen API key	Required for AI avatar integration (`ai-avatar-video`)
ElevenLabs API key	Optional for premium voiceover quality

Key Capabilities¶

Script-to-Video Pipeline¶

┌──────────────┐     ┌───────────────┐     ┌──────────────┐     ┌─────────────┐
│  Text/Outline │ ──▶ │  Scene Comp   │ ──▶ │  AV Render   │ ──▶ │  Export     │
│  → Full Script│     │  → Visual Plan│     │  → MP4/WebM  │     │  → Social   │
└──────────────┘     └───────────────┘     └──────────────┘     └─────────────┘

Core Commands¶

Command	Purpose	Example
`ai-video generate`	Generate video from text	`ai-video generate --script "product-demo.md" --style "corporate"`
`ai-video script`	Generate script only	`ai-video script --topic "How CorpusIQ saves 10hrs/week" --duration 60s`
`ai-video scene`	Design scene composition	`ai-video scene --script "demo.md" --scenes 5 --transitions "fade"`
`ai-video avatar`	Add AI presenter	`ai-video avatar --script "demo.md" --avatar "professional-female" --voice "sarah"`
`ai-video export`	Export final video	`ai-video export --format "mp4" --resolution "1080p" --platform "youtube"`

AI Avatars (`ai-avatar-video`)¶

# Generate a product demo with AI presenter
ai-video avatar \
  --script "corpusiq-overview.md" \
  --avatar "business-casual-male" \
  --voice "adam" \
  --background "modern-office" \
  --gestures "natural" \
  --output "corpusiq-demo.mp4"

Available avatar styles: professional, casual, creative, technical Voice options: 20+ voices across male/female, accents (US, UK, AU, IN) Background options: office, studio, gradient, transparent, custom-image

Scene Composition¶

# scene-plan.yaml
scenes:
  - duration: 10s
    type: intro
    content: "Logo animation + hook text"
    transition: fade-in
  - duration: 30s
    type: presenter
    content: "Problem statement - AI avatar on screen"
    avatar: business-casual-male
  - duration: 40s
    type: screen-recording
    content: "CorpusIQ dashboard walkthrough"
    overlay: presenter-pip
  - duration: 15s
    type: cta
    content: "Visit corpusiq.io - free trial"
    transition: zoom-out

Multi-Format Export¶

Format	Resolution	Best For
MP4 (H.264)	1080p / 4K	YouTube, website, download
MP4 (Vertical)	1080x1920	TikTok, Reels, Shorts
WebM	1080p	Web embedding
GIF	480p	Twitter, Reddit, email
Social-optimized	Platform-specific	Auto-detects from `--platform` flag

Common Workflows for CorpusIQ¶

Daily UGC Video (60s Product Tip)¶

# Generate script → avatar video → social export
ai-video script \
  --topic "How to find cash flow leaks with CorpusIQ" \
  --tone "technical,helpful" \
  --duration 60 \
  --output "daily-tip-script.md"

ai-video avatar \
  --script "daily-tip-script.md" \
  --avatar "technical-expert" \
  --voice "marcus" \
  --output "daily-tip-$(date +%Y%m%d).mp4"

ai-video export \
  --input "daily-tip-$(date +%Y%m%d).mp4" \
  --platforms "tiktok,reels,youtube-shorts"

Product Launch Video (Multi-Scene)¶

# Generate complete launch video with presenter + screen recording
ai-video generate \
  --script "launch/corpusiq-v2-announcement.md" \
  --style "product-launch" \
  --avatar "professional-female" \
  --screenshots "launch/screenshots/" \
  --music "upbeat-corporate" \
  --duration 120 \
  --output "corpusiq-v2-launch.mp4"

Blog-to-Video Repurposing¶

# Convert blog post to short-form video
ai-video generate \
  --source "https://corpusiq.io/blog/ai-for-operators" \
  --format "blog-to-video" \
  --duration 90 \
  --platforms "youtube,linkedin" \
  --output "ai-for-operators-video.mp4"

Integration with CorpusIQ Video Stack¶

Tool	Role	When to Use
ai-video-generation (this)	Script-to-video pipeline	Full AI video production
HyperFrames (108K)	Template-based video	Product demos, website-to-video
Remotion (430K)	Programmatic video (React)	Custom animations, data-driven video
HeyGen API	Avatar rendering	Underlying engine for ai-avatar-video
ElevenLabs	Premium voiceover	When default TTS quality insufficient

Recommendation: Use ai-video-generation for daily content (fast, automated). Use HyperFrames for product launch videos (template-driven, polished). Use Remotion for data-driven videos (custom logic, charts).

Troubleshooting¶

Symptom	Cause	Fix
`ai-video: command not found`	Skill not installed	`npx skills add 101-skills/skills@ai-video-generation`
Avatar lip-sync misaligned	Audio/video timing drift	Add `--sync-mode "precise"` (slower but accurate)
Rendering slow (5+ min/video)	CPU rendering	Enable GPU acceleration or reduce resolution to 720p
Voice sounds robotic	Default TTS engine	Install ElevenLabs and pass `--voice-provider "elevenlabs"`
Export fails on TikTok format	Missing vertical template	Use `--platform "tiktok"` for auto-cropping
FFmpeg errors	Missing or outdated FFmpeg	`sudo apt update && sudo apt install ffmpeg`

Integration with CorpusIQ¶

CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.

This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.

See what CorpusIQ connects