| name | remotion-ecosystem |
| description | Unified skill ecosystem for Remotion programmatic video production at scale. Orchestrates 91 skills across 10 categories: core video rendering, data ingestion, browser automation, AI intelligence, content design, monitoring, marketing strategy, devops deployment, document output, and agent systems. Use when building end-to-end video pipelines that combine data sourcing, AI generation, rendering, A/B testing, deployment, and monitoring into a cohesive production system.
|
| requires | Node.js 18+, React/TypeScript, z-ai-web-dev-sdk |
| integrates | ["remotion-video-pro (core rendering engine)","firecrawl-scrape (content acquisition)","browser-cdp (auth session reuse)","marketing-video (video strategy)","marketing-ab-testing (video variant testing)","amazon-bedrock (AI model orchestration)","sentry-ai-monitoring (LLM call tracing)","think (structured reasoning)","council-of-five (multi-perspective debate)"] |
| always_load | ["_ARCHITECTURE.md","_PIPELINE-ROUTES.md"] |
| conditional_load | {"data_source_needed":["02-data-ingestion/SKILL.md"],"browser_auth_needed":["03-browser-automation/SKILL.md"],"ai_generation_needed":["04-ai-intelligence/SKILL.md"],"deployment_needed":["08-devops-deploy/SKILL.md"]} |
Remotion Ecosystem — Unified Skill Architecture
Scope
This is the integration layer that transforms 91 individual skills into a
cohesive video production pipeline. It does not replace any skill — it provides
the routing, data flow, and orchestration patterns that connect them.
Every skill below can be used independently. This ecosystem document describes
how they compose together when building end-to-end Remotion video systems.
The Ten Categories
remotion-ecosystem/
├── 01-core-video/ ← Rendering engine + video strategy
│ ├── remotion-video-pro (scaffolding, animation, batch render)
│ ├── marketing-video (production approach selection)
│ ├── video-generation (AI video creation)
│ └── video-understand (video analysis)
│
├── 02-data-ingestion/ ← Content sourcing for video data
│ ├── firecrawl-scrape (JS-rendered page extraction)
│ ├── firecrawl-crawl (bulk site crawling)
│ ├── spider (site scanning + security audit)
│ ├── web-reader (page metadata + content)
│ ├── web-search (real-time search)
│ └── browserbase-fetch (lightweight URL fetching)
│
├── 03-browser-automation/ ← Auth + interaction for protected content
│ ├── browser-cdp (Chrome session reuse)
│ ├── agent-browser (headful browser automation)
│ └── browserbase-autobrowse (self-improving browse loops)
│
├── 04-ai-intelligence/ ← AI reasoning + model orchestration
│ ├── think (multi-framework reasoning via Gemini)
│ ├── council-of-five (5-persona parallel debate)
│ ├── amazon-bedrock (Claude/Llama/Nova model invocation)
│ ├── LLM (chat completions)
│ └── VLM (vision-language understanding)
│
├── 05-content-design/ ← Visual assets + motion design
│ ├── charts (data visualization)
│ ├── image-generation (AI image creation)
│ ├── image-edit (AI image editing)
│ ├── ui-ux-pro-max (design system + UI reasoning)
│ ├── anthropic-frontend-design (distinctive interfaces)
│ └── web-shader-extractor (visual effect extraction)
│
├── 06-monitoring-security/ ← Observability + protection
│ ├── sentry-ai-monitoring (LLM call tracing + token tracking)
│ ├── sentry-nextjs-sdk (Next.js error + perf monitoring)
│ ├── audit-analyzer (system auditing)
│ ├── diagnose (debugging + diagnosis)
│ └── mcp-spy (MCP traffic debugging)
│
├── 07-marketing-strategy/ ← Growth + experimentation
│ ├── marketing-ab-testing (hypothesis + chi-square significance)
│ ├── marketing-launch (5-phase launch framework)
│ ├── marketing-mode (growth marketing orchestration)
│ └── content-strategy (content planning)
│
├── 08-devops-deploy/ ← Infrastructure + CI/CD
│ ├── aws-agents-deploy (Railway/Vercel/Lambda deployment)
│ ├── amazon-bedrock (AWS AI model deployment)
│ ├── vercel (Vercel deployment)
│ ├── deployment-manager (multi-platform deploy)
│ ├── github (CI/CD + repo management)
│ └── fullstack-dev (Next.js full-stack development)
│
├── 09-document-output/ ← Reports + deliverables from video data
│ ├── docx (Word documents)
│ ├── pdf (PDF generation)
│ ├── ppt / pptx (presentations)
│ ├── xlsx (spreadsheets)
│ └── infographic-gen (infographics)
│
└── 10-agent-systems/ ← Agent orchestration + memory
├── agent-roles (specialized agent personas)
├── superpowers (spec-first + sub-agent delegation)
├── proactive-agent (autonomous self-improvement)
├── self-improving-agent (error logging + learning)
├── persistent-memory (cross-session memory)
└── context-compressor (context window optimization)
Pipeline Architecture
The Remotion ecosystem follows a 6-stage pipeline:
[1. ACQUIRE] → [2. THINK] → [3. DESIGN] → [4. RENDER] → [5. TEST] → [6. DEPLOY]
┌──────────────────────────────────────────────────────────────────┐
│ Stage 1: ACQUIRE — Data Ingestion (02-data-ingestion) │
│ │
│ firecrawl-scrape ──→ scrape competitor pages │
│ browser-cdp ──────→ extract behind-auth content │
│ spider ───────────→ scan product pages for data │
│ web-search ────────→ gather market intelligence │
│ web-reader ────────→ extract article content │
│ │
│ Output: { products, competitors, testimonials, pricing } │
└────────────────────────────┬─────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────────┐
│ Stage 2: THINK — AI Intelligence (04-ai-intelligence) │
│ │
│ think ─────────────→ first-principles analysis of data │
│ council-of-five ───→ 5-persona debate on video approach │
│ amazon-bedrock ────→ generate script / copy / storyboard │
│ LLM ────────────────→ refine messaging and CTAs │
│ │
│ Output: { script, storyboard, copy variants, scene plan } │
└────────────────────────────┬─────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────────┐
│ Stage 3: DESIGN — Content + Visual (05-content-design) │
│ │
│ image-generation ──→ hero images, product shots │
│ charts ─────────────→ data visualizations for video │
│ ui-ux-pro-max ─────→ layout + color + typography system │
│ web-shader-extractor → visual effects from reference sites │
│ │
│ Output: { images, charts, color palettes, motion specs } │
└────────────────────────────┬─────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────────┐
│ Stage 4: RENDER — Core Video (01-core-video) │
│ │
│ remotion-video-pro → compose React components as frames │
│ marketing-video ────→ select production approach │
│ video-generation ──→ AI-generated B-roll footage │
│ video-understand ──→ analyze reference videos │
│ │
│ Output: { MP4 renders, variant compositions } │
└────────────────────────────┬─────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────────┐
│ Stage 5: TEST — Monitoring + A/B (06 + 07) │
│ │
│ marketing-ab-testing → chi-square significance on variants │
│ sentry-ai-monitoring → trace LLM calls during generation │
│ audit-analyzer ─────→ security scan of rendered assets │
│ diagnose ────────────→ debug rendering failures │
│ │
│ Output: { test results, significance scores, error reports } │
└────────────────────────────┬─────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────────┐
│ Stage 6: DEPLOY — DevOps + Marketing (08 + 07) │
│ │
│ aws-agents-deploy ─→ deploy to Lambda/Railway │
│ marketing-launch ───→ 5-phase product launch │
│ vercel ──────────────→ webhook service hosting │
│ fullstack-dev ───────→ Next.js dashboard for render management │
│ │
│ Output: { live renders, webhook endpoints, launch campaign } │
└──────────────────────────────────────────────────────────────────┘
Integration Routes
Route A: Competitor Intelligence Video
spider (scan competitor sites)
→ firecrawl-scrape (extract product pages)
→ amazon-bedrock (analyze competitive landscape)
→ council-of-five (debate video angle)
→ charts (create comparison visualizations)
→ remotion-video-pro (render competitor comparison video)
→ marketing-ab-testing (A/B test video variants)
→ aws-agents-deploy (deploy to Lambda)
Route B: Product Launch Video Campaign
marketing-launch (plan 5-phase launch)
→ web-search (gather market data)
→ think (first-principles messaging strategy)
→ amazon-bedrock (generate video script)
→ image-generation (create hero images)
→ ui-ux-pro-max (design visual system)
→ remotion-video-pro (render launch video)
→ video-generation (AI B-roll footage)
→ sentry-ai-monitoring (trace all AI calls)
→ aws-agents-deploy (deploy pipeline)
→ marketing-ab-testing (test hook variations)
Route C: Data-Driven Personalized Videos
firecrawl-crawl (bulk extract customer data sources)
→ browser-cdp (extract behind-auth CRM data)
→ amazon-bedrock (personalize video copy per recipient)
→ charts (generate personalized data visualizations)
→ remotion-video-pro (batch render with personalization props)
→ sentry-ai-monitoring (monitor batch LLM token usage)
→ aws-agents-deploy (Lambda batch rendering)
→ docx/xlsx (generate ROI report from campaign)
Route D: Content Repurposing Pipeline
video-understand (analyze source video)
→ browserbase-autobrowse (gather reference content)
→ think (plan repurposing strategy)
→ firecrawl-scrape (extract blog content for reference)
→ marketing-video (select short-form vs long-form approach)
→ image-generation (create thumbnail variants)
→ remotion-video-pro (render platform-specific versions)
→ marketing-launch (coordinate multi-platform distribution)
Route E: A/B Test Video Variants at Scale
marketing-ab-testing (design hypothesis + variants)
→ council-of-five (debate which variants to test)
→ amazon-bedrock (generate variant copy/scripts)
→ ui-ux-pro-max (design variant visual systems)
→ remotion-video-pro (render A/B compositions)
→ sentry-ai-monitoring (trace generation costs)
→ marketing-launch (deploy test to audience)
→ charts (visualize test results)
→ pdf (generate test report)
Cross-Cutting Concerns
Security (06-monitoring-security)
Every pipeline stage should be wrapped with:
- sentry-ai-monitoring: Trace all LLM calls (token count, latency, cost)
- sentry-nextjs-sdk: Monitor webhook service health
- audit-analyzer: Periodic security scan of deployed infrastructure
- mcp-spy: Debug MCP tool integrations in development
Memory (10-agent-systems)
Cross-session continuity for recurring video production:
- persistent-memory: Store brand guidelines, color palettes, past campaigns
- context-compressor: Optimize context window when running complex pipelines
- self-improving-agent: Learn from rendering failures and A/B test results
Quality (01-core-video)
Enforce remotion-video-pro golden rules at every render:
- Frame-based determinism (no Math.random)
- Spring animations over linear
- React.memo on every scene component
- Zod schema validation on all props
- Pre-commit validation hooks
Quick Reference — Skill to Pipeline Stage
| Skill | Primary Stage | Secondary Use |
|---|
| remotion-video-pro | 4. RENDER | 5. TEST (validation) |
| firecrawl-scrape | 1. ACQUIRE | 6. DEPLOY (verify pages) |
| firecrawl-crawl | 1. ACQUIRE | 3. DESIGN (reference content) |
| spider | 1. ACQUIRE | 6. DEPLOY (security scan) |
| browser-cdp | 1. ACQUIRE | 3. DESIGN (auth content) |
| browserbase-fetch | 1. ACQUIRE | — |
| browserbase-autobrowse | 1. ACQUIRE | 2. THINK (research) |
| think | 2. THINK | 5. TEST (debug analysis) |
| council-of-five | 2. THINK | 7. MARKETING (strategy) |
| amazon-bedrock | 2. THINK | 4. RENDER (AI generation) |
| LLM | 2. THINK | — |
| VLM | 2. THINK | 4. RENDER (scene analysis) |
| charts | 3. DESIGN | 9. OUTPUT (reports) |
| image-generation | 3. DESIGN | 4. RENDER (assets) |
| ui-ux-pro-max | 3. DESIGN | 2. THINK (layout reasoning) |
| marketing-ab-testing | 5. TEST | 7. MARKETING (growth) |
| marketing-launch | 6. DEPLOY | 7. MARKETING (campaign) |
| sentry-ai-monitoring | 5. TEST | 6. DEPLOY (production monitor) |
| aws-agents-deploy | 6. DEPLOY | — |
| fullstack-dev | 6. DEPLOY | 3. DESIGN (dashboards) |
| docx/pdf/xlsx | 9. OUTPUT | 7. MARKETING (reports) |
| agent-roles | 10. SYSTEMS | 2. THINK (delegation) |
| superpowers | 10. SYSTEMS | — |
