Youtube channel parser

The problem

Manually transcribing and summarizing YouTube content is slow, especially when dealing with entire channels or specific date ranges. Developers often lack a reliable way to filter videos by metadata or speakers before triggering expensive transcription workflows.

What it does

• Inventories entire YouTube channels and filters videos by date, topic, speaker, or metadata.
• Retrieves direct transcripts or subtitles to minimize processing time.
• Falls back to local audio download and Whisper transcription when official transcripts are unavailable.
• Generates structured JSON, CSV, and Markdown reports for downstream analysis.
• Handles bot-detection and environment-specific network restrictions gracefully.

Frameworks & tools

Built with yt-dlp, youtube-transcript-api, and faster-whisper. Orchestrated via Python scripts for high-performance metadata normalization and transcription fallbacks.

Why this beats prompting it yourself

Standard LLM prompts cannot access video metadata or audio files directly without a specialized execution environment. This skill manages the end-to-end pipeline, including multi-stage filtering to save tokens and local transcription to bypass API limits.

Use cases

• Extract all video transcripts from a specific creator within a date window for RAG pipelines.
• Generate a summary report of every video on a channel that mentions a specific topic.
• Automate the creation of structured show notes and JSON metadata for video archives.
• Audit a competitor channel by exporting their entire video inventory to CSV.

Known limitations

Requires environment access for network requests and audio processing. Subject to YouTube content-access restrictions and bot-detection limits.