` — Video title - `<link>` — YouTube URL (format: `https://www.youtube.com/watch?v=VIDEO_ID` or `/shorts/VIDEO_ID`) - `<published>` — Publication date - `<media:description>` — Full description with chapters, links, and speaker info Parse both feeds and build a combined video inventory. ### Step 2: Filter Videos **INCLUDE** videos that are: - Published within the newsletter's DATE_RANGE (or up to 2 weeks before/after) - About GitHub Copilot, VS Code, agents, MCP, security, or enterprise features - Full-length videos (>2 minutes) or Shorts that directly demonstrate a named feature **EXCLUDE**: - Shorts that are generic teasers without specific feature content - Open Source Friday episodes about non-GitHub projects - Non-English content - General tutorials not tied to a specific feature in the newsletter ### Step 3: Match Videos to Newsletter Entries For each video, attempt to match it to a newsletter entry using these signals (in priority order): 1. **Title keyword match**: Video title contains a key phrase from the newsletter entry (e.g., "Subagents" matches "Parallel Subagents (`GA`)") 2. **Description link match**: Video description links to the same changelog/release-notes URL as the newsletter entry 3. **Topic overlap**: Video covers the same feature area (e.g., a video about "MCP Apps" matches the "MCP Ecosystem Expansion" bullet) **Confidence levels**: - **HIGH (90%+)**: Title directly references the feature name OR description links to the same changelog URL. Add the link. - **MEDIUM (60-80%)**: Topic overlap but not a direct title/URL match. Add only if no HIGH match exists for that entry. - **LOW (<60%)**: Tangential connection. Do NOT add. ### Step 4: Estimate Video Duration The RSS feed does NOT include video duration. Estimate using these heuristics: - **Shorts** (`/shorts/` in URL): Label as `(1m)` - **VS Code Live / livestream**: Label as `(60m)` (typically 1-2 hours, but 60m is a safe estimate) - **Podcast episodes**: Label as `(25m)` - **Regular videos**: Label as `(5m)` unless description chapters suggest longer (sum chapter timestamps) - If the description contains chapter timestamps (e.g., "14:56 In summary"), use the last timestamp rounded up to estimate duration ### Step 5: Add Video Links For each matched entry, append a `[Video (Xm)](URL)` link. Place it alongside existing links using the standard pipe separator: **Before**: ```markdown - **Parallel Subagents (`GA`)** -- Description. - [Release Notes](URL) | [Docs](URL) ``` **After**: ```markdown - **Parallel Subagents (`GA`)** -- Description. - [Release Notes](URL) | [Docs](URL) | [Video (25m)](https://www.youtube.com/watch?v=GMAoTeD9siU) ``` For inline-linked entries (7+ links), add the video link at the end of the "See also:" line or as the last inline link. ### Step 6: Validate - [ ] All added video links are valid YouTube URLs - [ ] Video durations are reasonable estimates - [ ] No duplicate video links (same video linked to multiple entries) - [ ] Only HIGH confidence matches were added - [ ] Links use the format `[Video (Xm)](URL)` ## Reference - [Video Sources](references/video-sources.md) — Channel IDs, RSS feed URLs, matching heuristics - [Content Format Spec](../content-curation/references/content-format-spec.md) — Link format rules ## Done When - [ ] Both RSS feeds fetched and parsed - [ ] Videos matched to newsletter entries (HIGH confidence only) - [ ] `[Video (Xm)]` links added to matched entries - [ ] All links validated - [ ] Enriched newsletter written to disk --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/briancl2) — claim your Tome and manage your conversions. <!-- tomevault:4.0:skill

# Briancl2 Customernewsletter Video Matching > Video Matching - Skill: `tomevault-io/briancl2-customernewsletter-video-matching` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/briancl2-customernewsletter-video-matching` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/briancl2-customernewsletter-video-matching/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/briancl2-customernewsletter-video-matching --- # Video Matching Enrich newsletter entries with links to matching official YouTube videos from the VS Code and GitHub channels. ## Quick Start 1. Read the assembled newsletter from `output/YYYY-MM_month_newsletter.md` 2. Fetch RSS feeds from VS Code and GitHub YouTube channels 3. Match videos to newsletter entries by topic 4. Add `[Video (Xm)](URL)` links to matched entries 5. Write enriched newsletter back to `output/YYYY-MM_month_newsletter.md` ## Inputs - **Newsletter**: `output/YYYY-MM_month_newsletter.md` (required) - **YouTube RSS feeds** (fetch these): - VS Code channel: `https://www.youtube.com/feeds/videos.xml?channel_id=UCs5Y5_7XK8HLDX0SLNwkd3w` - GitHub channel: `https://www.youtube.com/feeds/videos.xml?channel_id=UC7c3Kb6jYCRj4JOHHZTxKsQ` ## Output - **Enriched Newsletter**: `output/YYYY-MM_month_newsletter.md` (overwrite in place) - Video matching report (inline with work) ## Core Workflow ### Step 1: Fetch Video Feeds Fetch both YouTube RSS feeds. Each returns the 15 most recent videos with: - `` — Video title - `<link>` — YouTube URL (format: `https://www.youtube.com/watch?v=VIDEO_ID` or `/shorts/VIDEO_ID`) - `<published>` — Publication date - `<media:description>` — Full description with chapters, links, and speaker info Parse both feeds and build a combined video inventory. ### Step 2: Filter Videos **INCLUDE** videos that are: - Published within the newsletter's DATE_RANGE (or up to 2 weeks before/after) - About GitHub Copilot, VS Code, agents, MCP, security, or enterprise features - Full-length videos (>2 minutes) or Shorts that directly demonstrate a named feature **EXCLUDE**: - Shorts that are generic teasers without specific feature content - Open Source Friday episodes about non-GitHub projects - Non-English content - General tutorials not tied to a specific feature in the newsletter ### Step 3: Match Videos to Newsletter Entries For each video, attempt to match it to a newsletter entry using these signals (in priority order): 1. **Title keyword match**: Video title contains a key phrase from the newsletter entry (e.g., "Subagents" matches "Parallel Subagents (`GA`)") 2. **Description link match**: Video description links to the same changelog/release-notes URL as the newsletter entry 3. **Topic overlap**: Video covers the same feature area (e.g., a video about "MCP Apps" matches the "MCP Ecosystem Expansion" bullet) **Confidence levels**: - **HIGH (90%+)**: Title directly references the feature name OR description links to the same changelog URL. Add the link. - **MEDIUM (60-80%)**: Topic overlap but not a direct title/URL match. Add only if no HIGH match exists for that entry. - **LOW (<60%)**: Tangential connection. Do NOT add. ### Step 4: Estimate Video Duration The RSS feed does NOT include video duration. Estimate using these heuristics: - **Shorts** (`/shorts/` in URL): Label as `(1m)` - **VS Code Live / livestream**: Label as `(60m)` (typically 1-2 hours, but 60m is a safe estimate) - **Podcast episodes**: Label as `(25m)` - **Regular videos**: Label as `(5m)` unless description chapters suggest longer (sum chapter timestamps) - If the description contains chapter timestamps (e.g., "14:56 In summary"), use the last timestamp rounded up to estimate duration ### Step 5: Add Video Links For each matched entry, append a `[Video (Xm)](URL)` link. Place it alongside existing links using the standard pipe separator: **Before**: ```markdown - **Parallel Subagents (`GA`)** -- Description. - [Release Notes](URL) | [Docs](URL) ``` **After**: ```markdown - **Parallel Subagents (`GA`)** -- Description. - [Release Notes](URL) | [Docs](URL) | [Video (25m)](https://www.youtube.com/watch?v=GMAoTeD9siU) ``` For inline-linked entries (7+ links), add the video link at the end of the "See also:" line or as the last inline link. ### Step 6: Validate - [ ] All added video links are valid YouTube URLs - [ ] Video durations are reasonable estimates - [ ] No duplicate video links (same video linked to multiple entries) - [ ] Only HIGH confidence matches were added - [ ] Links use the format `[Video (Xm)](URL)` ## Reference - [Video Sources](references/video-sources.md) — Channel IDs, RSS feed URLs, matching heuristics - [Content Format Spec](../content-curation/references/content-format-spec.md) — Link format rules ## Done When - [ ] Both RSS feeds fetched and parsed - [ ] Videos matched to newsletter entries (HIGH confidence only) - [ ] `[Video (Xm)]` links added to matched entries - [ ] All links validated - [ ] Enriched newsletter written to disk --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/briancl2) — claim your Tome and manage your conversions.