Automated intelligence pipeline that monitors the AI/automation niche on YouTube, surfaces trends and competitor insights, and delivers a professional slide deck report.
- Collect — Searches 14 queries via YouTube Data API v3, deduplicates, fetches video details and channel stats
- Analyze — Extracts trends, calculates engagement benchmarks, ranks competitors, detects content gaps
- AI Insights (optional) — Generates video summaries, title hook patterns, and thumbnail analysis via Gemini
- Report — Builds a 12–15 slide PPTX deck (Deep Purple + Emerald design)
- Deliver — Sends styled HTML email with the PPTX attached
npm installCreate a .env file:
YOUTUBE_API_KEY=your_key
REPORT_EMAIL=recipient@example.com
GMAIL_USER=your_gmail@gmail.com
GMAIL_APP_PASSWORD=your_app_password
GEMINI_API_KEY=your_key # optional, for AI Insights
claude "/yt-report"This runs the full pipeline interactively, including the optional AI Insights stage.
node scripts/pipeline.jsThe pipeline runs automatically every Monday at 6:00 AM Israel time via Modal.
# Deploy the scheduled app
modal deploy modal_app.py
# Manual run
modal run modal_app.py::run_pipeline_manual
# Manual run without email
modal run modal_app.py::run_pipeline_manual --no-emailscripts/
pipeline.js # Main orchestrator
collect/youtube-api.js # YouTube Data API collector
analyze/analyze-data.js # Analysis orchestrator (trends, benchmarks, competitors, gaps)
report/generate-slides.js # PPTX deck generator
deliver/send-report.js # Email delivery
context/
search-config.json # 14 search queries + settings
modal_app.py # Modal deployment with weekly cron
YouTube Data API has a 10,000 unit/day limit. Each pipeline run uses ~1,400 units, allowing ~7 runs per day.
Use the yt-niche-pipeline skill to reconfigure search queries, trend terms, and slide branding for a different topic.
Private project.