Initial commit

commands/yt-transcribe.md

@@ -0,0 +1,59 @@
+---
+model: claude-sonnet-4-5
+allowed-tools: Task, Bash, Read, Write, WebFetch
+argument-hint: <video-url> [--output-format format] [--summary] [--language lang]
+description: Transcribe a YouTube video and extract key insights
+---
+
+# YouTube Transcribe Command
+
+Transcribe a YouTube video, fetch the transcript, and extract key insights and summaries.
+
+## Arguments
+
+**$1 (Required)**: `video-url`
+
+The full YouTube URL to transcribe. Example: `https://www.youtube.com/watch?v=dQw4w9WgXcQ`
+
+**Optional flags**:
+
+- `--output-format` - Output format: `text` (default), `markdown`, `json`, `vtt`
+- `--summary` - Generate a concise summary alongside full transcript
+- `--language` - Specific language code (e.g., `en`, `es`, `fr`). Default: auto-detect
+- `--timestamps` - Include timestamps in output
+- `--speakers` - Identify and separate speaker segments
+
+## Examples
+
+```bash
+/yt-transcribe "https://www.youtube.com/watch?v=VIDEO_ID"
+/yt-transcribe "https://www.youtube.com/watch?v=VIDEO_ID" --summary --format markdown
+/yt-transcribe "https://www.youtube.com/watch?v=VIDEO_ID" --language es --timestamps
+/yt-transcribe "https://www.youtube.com/watch?v=VIDEO_ID" --summary --speakers
+```
+
+## How It Works
+
+1. **Video Analysis** - Uses the video-analyzer agent to inspect the video
+2. **Transcript Fetching** - Uses the transcript-fetcher agent to get the best available transcript
+3. **Processing** - Uses the transcript-processor agent to format and optionally summarize
+4. **Output** - Delivers transcript in your specified format
+
+## Output
+
+The command returns:
+- Complete transcript with timestamps (if requested)
+- Speaker identification and segmentation (if requested)
+- Summary and key points (if requested)
+- Metadata including duration, language, source quality
+
+## Tips
+
+- For videos without captions, the system will attempt speech-to-text transcription
+- Add `--summary` to get a concise overview alongside the full transcript
+- Use `--speakers` to identify who's speaking and when
+- Markdown format is best for reading; JSON is best for downstream processing
+
+---
+
+Invoke the transcript-fetcher agent to retrieve the transcript, then use the transcript-processor agent to format and analyze it.