YouTube Daily Digest

SKILL.md

---
name: youtube-daily-digest
description: "Create a daily YouTube digest task that monitors specified channels, fetches new videos from the last 24 hours, retrieves transcripts via Supadata API, generates AI summaries, and sends a professional HTML email digest. Triggers: 'youtube daily digest', 'youtube summary task', 'channel digest', 'video digest email'."
metadata:
  openclaw:
    requires:
      env:
        - SUPADATA_API_KEY
---

# YouTube Daily Digest Skill

Create a daily automated task that monitors YouTube channels, fetches transcripts, summarizes videos, and delivers a professional HTML email digest.

## Prerequisites

**Required: Supadata API Key**
This skill requires a Supadata API key for fetching YouTube video transcripts.
- Sign up at https://dash.supadata.ai
- Get your API key from the dashboard
- Store it as `SUPADATA_API_KEY` environment variable for the task

Supadata provides reliable YouTube transcript extraction without IP blocking issues that affect cloud environments.

## Usage

When the user wants to set up a daily YouTube digest:

1. Ask for the YouTube channel URLs or handles they want to monitor (e.g., `@lexfridman`, `@allin`)
2. Ask for the recipient email address for the digest
3. Ask for preferred time (default: 8:00 AM in user's timezone)
4. Create a recurring task with this skill's workflow

## Task Setup Workflow

### Step 1: Resolve Channel IDs
For each channel URL/handle provided, resolve to a YouTube channel ID and RSS feed URL:

```python
import requests
import re

def get_channel_rss(channel_url):
    # Fetch channel page to extract channel ID
    r = requests.get(channel_url, headers={"User-Agent": "Mozilla/5.0"}, timeout=10)
    # Try multiple patterns to find channel ID
    patterns = [
        r'"channelId":"(UC[^"]+)"',
        r'"externalId":"(UC[^"]+)"',
    ]
    for pattern in patterns:
        match = re.search(pattern, r.text)
        if match:
            channel_id = match.group(1)
            return {
                "channel_id": channel_id,
                "rss_url": f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
            }
    raise ValueError(f"Could not find channel ID for {channel_url}")
```

### Step 2: Create Task Description
Create a task with the following description template, substituting the resolved channels:

---
**Task Description Template:**

Fetch new YouTube videos from the specified channels posted in the last 24 hours, get their transcripts via Supadata API, summarize each one, and send a daily email digest.

## Channels to Monitor
[List resolved channels with their channel_id and rss_url]

## Execution Steps

### 1. Fetch new videos from RSS feeds
For each channel, fetch its RSS feed XML and parse entries published within the last 24 hours.

```python
import requests
import xml.etree.ElementTree as ET
from datetime import datetime, timezone, timedelta

def fetch_new_videos(channels):
    cutoff = datetime.now(timezone.utc) - timedelta(hours=24)
    new_videos = []
    
    for channel_name, channel_id in channels:
        rss_url = f"https://www.youtube.com/feeds/videos.xml?channel_id={channel_id}"
        r = requests.get(rss_url, timeout=10)
        root = ET.fromstring(r.content)
        ns = {"atom": "http://www.w3.org/2005/Atom"}
        
        for entry in root.findall("atom:entry", ns):
            published = entry.find("atom:published", ns).text
            pub_dt = datetime.fromisoformat(published.replace("Z", "+00:00"))
            if pub_dt >= cutoff:
                title = entry.find("atom:title", ns).text
                vid_id = entry.find("atom:id", ns).text.split(":")[-1]
                new_videos.append({
                    "channel": channel_name,
                    "title": title,
                    "video_id": vid_id,
                    "url": f"https://youtu.be/{vid_id}",
                    "published": published[:10]
                })
    return new_videos
```

### 2. Fetch transcripts via Supadata API
```python
import os

api_key = os.environ.get("SUPADATA_API_KEY")

for video in new_videos:
    resp = requests.get(
        "https://api.supadata.ai/v1/youtube/transcript",
        headers={"x-api-key": api_key},
        params={"videoId": video["video_id"], "text": "true"},
        timeout=30
    )
    data = resp.json()
    video["transcript"] = data.get("content", "")
```

### 3. Summarize each video
For each video with a non-empty transcript, generate a concise summary (5-8 bullet points) covering key topics and insights. If transcript is empty, note "Transcript not available."

### 4. Send professional HTML email digest
**MUST USE HTML FORMAT** - Never send plain text email for this digest.

Send to the configured recipient email with:
- Subject: `YouTube Digest — {Month Day, Year}`
- body_format: `html`
- Use the professional HTML template provided in references/email-template.html

If no new videos found, send brief notification: "No new videos from your subscriptions in the last 24 hours."

## Environment Variables
- SUPADATA_API_KEY: Supadata API key for transcript fetching (required)
---

### Step 3: Schedule the Task
Create a recurring task with schedule (default: "Every day at 8AM" in user's timezone).
Use handler_profile "taskworker".

### Step 4: Store API Key
Ensure SUPADATA_API_KEY is configured for the task/skill.

## HTML Email Template Reference
See references/email-template.html for the complete professional HTML email template to use.

## Notes
- YouTube RSS feeds return max 15 recent videos per channel
- Supadata API has rate limits based on your plan (check https://supadata.ai/pricing)
- Video transcripts may not be available for all videos (private, age-restricted, no captions)
- Task agent should handle empty transcript gracefully