GDC URL to VTT Skill (Parent)

Purpose

This is the parent skill for gdc-vtt-to-note. It handles the case where the user provides a GDC Vault video page URL rather than a raw VTT segment URL.

It automatically:

1. Authenticates with GDC Vault
2. Parses the video page to extract the Blazestreaming video ID
3. Follows the CDN chain to locate a VTT caption segment URL
4. Delegates to the gdc-vtt-to-note workflow to capture, summarize, and create an Obsidian note

Trigger

• User provides a GDC Vault video page URL, e.g.:
• https://gdcvault.com/play/1035765/Artists-Do-You-Want-to
• Any URL matching gdcvault.com/play/*
• Do NOT trigger when the user provides a raw .vtt URL — that is handled by gdc-vtt-to-note directly.

Required Input

• A GDC Vault video page URL (the page you watch the video on, NOT a VTT file URL)
• GDC Vault credentials (email + password)

Credential Resolution

Credentials are needed for Step 1. Look for them in this order:

1. Read the note literature notes/video notes/gdc/GDC password.md for the latest username/password
2. If not found, ask the user to provide --email and --password

Workflow

Step 1: Extract VTT URL

Run the bundled extraction script with the user's video URL and credentials:

python3 .claude/skills/gdc-url-to-vtt/scripts/extract_vtt_url.py \
  "<gdc_video_url>" \
  --email "<email>" \
  --password "<password>"

The script:

• Logs into GDC Vault (cookies are saved to .claude/skills/gdc-url-to-vtt/scripts/.gdc_cookies.txt for reuse)
• Extracts the Blazestreaming videoId from the page
• Follows the HLS manifest chain to find the first VTT caption segment
• Prints the full VTT URL to stdout

If extraction fails:

• Report the error to the user
• Common failures: expired credentials, page not found, no subtitle track
• Do NOT proceed to Step 2

Step 2: Capture Captions (delegate to gdc-vtt-to-note)

Once the VTT URL is obtained, run the capture script from gdc-vtt-to-note:

python3 .claude/skills/gdc-vtt-to-note/scripts/capture_gdc_vtt.py \
  "<vtt_url_from_step_1>" \
  --output captions.txt

This brute-forces all VTT chunk IDs to assemble the full caption stream.

If the script fails, report the error and do NOT proceed.

Step 3: Summarize & Create Note (follow gdc-vtt-to-note post-processing)

Once captions.txt is generated:

1. Read the generated caption file
2. Build a concise summary including:

• Topic
• Key points
• Actionable insights

1. Create an Obsidian note in literature notes/video notes/gdc/ai notes

Note Naming Rule

• Filename format: YYYY-MM-DD-{summary_slug}.md
• Slug rules: lowercase, spaces → -, remove invalid filename characters
• If conflict exists, append -2, -3, etc.

Note Content

Include:

• Title
• Source GDC video URL
• Resolved VTT URL
• Caption file path
• Summary sections
• Captions.txt full content as reference

Prefer frontmatter metadata:

• source_url: the original GDC video page URL
• source_vtt: the resolved VTT segment URL
• caption_file: path to captions.txt
• created_at: today's date
• tags: [gdc, summary]

Safety

• If Step 1 (VTT extraction) fails, report the error and stop
• If Step 2 (caption capture) fails, report stderr and do not create the note
• If caption file is empty, create note with incomplete status
• Never modify the gdc-vtt-to-note skill or its scripts
• Cookie jar file is stored alongside the script — add to .gitignore if needed