cc-dev-template 0.1.49 → 0.1.50

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "cc-dev-template",
-  "version": "0.1.49",
+  "version": "0.1.50",
   "description": "Structured AI-assisted development framework for Claude Code",
   "bin": {
     "cc-dev-template": "./bin/install.js"

package/src/skills/youtube-to-notes/SKILL.md

@@ -0,0 +1,11 @@
+---
+name: youtube-to-notes
+description: This skill should be used when the user asks to "transcribe this YouTube video", "turn this video into notes", "get me notes from this YouTube talk", "summarize this YouTube video", "convert this video to markdown", or says "youtube to notes". Activate when the user wants to download, transcribe, and summarize a YouTube video into structured Markdown documents.
+argument-hint: "[url or search query]"
+---
+
+# YouTube to Notes
+
+## What To Do Now
+
+Read `references/step-1-setup.md`.

package/src/skills/youtube-to-notes/references/step-1-setup.md

@@ -0,0 +1,15 @@
+# Setup
+
+Verify `yt-dlp` and `whisper` are installed.
+
+If either is missing, tell the user the exact install command (`brew install yt-dlp` or `brew install openai-whisper`) and stop.
+
+## Resolve the Video
+
+The user provides either a YouTube URL or a search query via `$ARGUMENTS`.
+
+**If it's a URL:** Store it and move on.
+
+**If it's a search query:** Use yt-dlp to search YouTube for the top 5 results. Display each result's title, channel, duration, and URL. Ask the user to pick one.
+
+Read `references/step-2-download-transcribe.md` when a video URL is confirmed.

package/src/skills/youtube-to-notes/references/step-2-download-transcribe.md

@@ -0,0 +1,13 @@
+# Download and Transcribe
+
+## Download Audio
+
+Use yt-dlp to extract audio as high-quality MP3 from the selected URL. Use the video title as the output filename.
+
+## Transcribe
+
+Run Whisper on the MP3 file using the `base` model. Output format: plain text, current directory.
+
+If the user requested higher accuracy, use `small` or `medium` instead — warn that these are significantly slower.
+
+Read `references/step-3-create-documents.md` when the `.txt` transcript file exists.

package/src/skills/youtube-to-notes/references/step-3-create-documents.md

@@ -0,0 +1,31 @@
+# Create Documents
+
+Read the raw `.txt` transcript from Whisper. Create two Markdown files in the current directory.
+
+## Transcript: `<video-title>-transcript.md`
+
+| Element | Instructions |
+|---------|-------------|
+| **YAML frontmatter** | `title`, `channel`, `url`, `date` (today), `duration` |
+| **Section headers** (`##`) | Insert where the topic shifts |
+| **Paragraphs** | Break at natural pause points into short, readable paragraphs |
+| **Speaker labels** | Label speakers in conversations when confident from context. Omit when unsure. |
+| **Code blocks** | Wrap any code mentioned in the video |
+| **Blockquotes** | Use for notable quotes |
+
+## Summary: `<video-title>-summary.md`
+
+| Section | Content |
+|---------|---------|
+| **YAML frontmatter** | Same fields as transcript, plus `type: summary` |
+| **Overview** | 2-3 sentence summary |
+| **Key Points** | 5-10 bullets of the most important ideas |
+| **Notable Quotes** | 3-5 direct quotes (include approximate timestamps if available from Whisper output) |
+| **Action Items / Takeaways** | Concrete recommendations or next steps from the video |
+| **Related Topics** | Keywords or topics for further research |
+
+## Cleanup
+
+Delete the intermediate `.mp3` and `.txt` files unless the user asked to keep them.
+
+Tell the user what files were created and their full paths.