youtube-top-hits

name	youtube-top-hits
description	Analyze a YouTube channel's most popular videos using the Apify YouTube Scraper (streamers/youtube-scraper). Use this skill whenever the user asks about a YouTuber's top videos, biggest hits, most viewed content, best performing uploads, viral videos, or channel analytics. Also trigger when the user provides a YouTube channel URL and wants to know what content performed best, or asks to compare video performance across a channel. Even casual mentions like 'what are their best videos' or 'which videos blew up' should trigger this skill.
license	MIT
metadata	{"author":"matt-brown"}

YouTube Top Hits Analyzer

This skill fetches video data from a YouTube channel via the Apify API and identifies the top-performing videos by view count.

How It Works

1. The user provides a YouTube channel URL (e.g., https://www.youtube.com/@KevinStratvert)
2. You write out and run the Python script below, which calls the recommended Apify scraper, YouTube Scraper (streamers/youtube-scraper)
3. The script filters videos to the requested time window, sorts by views, and returns the top hits

Prerequisites

This skill calls a paid third-party API (Apify), so each user must supply their own Apify API token — there is no shared or built-in key.

• Apify account + token (required): Create a free account at https://apify.com, then copy your token from Settings → API & Integrations. The free tier includes monthly credits that are enough for occasional channel scrapes.
• Set the token as an environment variable named APIFY_API_KEY before running:
  • macOS / Linux: export APIFY_API_KEY="your_token_here"
  • Windows (PowerShell): $env:APIFY_API_KEY="your_token_here"
  • To avoid setting it every session, add the export line to your shell profile (~/.zshrc, ~/.bashrc) or your agent's environment/secrets config.
• The script reads the token only from this environment variable. If it is missing, the script stops immediately and tells the user how to set it — it never falls back to any other key.
• requests library: Should be pre-installed. If not: pip install requests --break-system-packages

Recommended Scraper: YouTube Scraper (streamers/youtube-scraper)

This skill is built around Apify's YouTube Scraper (actor id streamers/youtube-scraper), a maintained, general-purpose YouTube actor that extracts video title, URL, view count, likes, comments, duration, and upload date from a channel, video, playlist, or search URL.

One-time setup:

1. Create a free account at https://console.apify.com/sign-up.
2. Open the actor at https://apify.com/streamers/youtube-scraper and click Try for free to add it to your account (this is required before the API will run it for you).
3. Copy your API token from Settings → API & Integrations and set it as APIFY_API_KEY (see Prerequisites above).

How the script uses it: it calls the actor's REST API with the channel URL as a startUrl, pulls up to --max-results videos, then does the time-window filtering and view-count sorting locally — so you don't need to configure any sort/date options in the Apify console. The script requests regular videos only (shorts and streams are set to 0).

Cost note: this is a pay-per-result actor (priced per 1,000 videos — see the actor's Pricing tab for the current rate), drawn from your own Apify credits. Keep --max-results modest to control spend.

Cheaper / alternative actors: for channel-only analysis, Apify's purpose-built Fast YouTube Channel Scraper is often less expensive. To swap, change the ACTOR_ID constant near the top of the script and verify the new actor returns compatible fields (viewCount, date, title, url); the parsing helpers already try several common field-name variants.

Important: Network Access

This skill makes outbound HTTP calls to api.apify.com. It works in environments with unrestricted network access (Claude Code, local terminal). It will NOT work in sandboxed environments that block external API calls.

Usage

Save the script below as fetch_top_hits.py (a temp directory is fine), confirm APIFY_API_KEY is set, then run:

python fetch_top_hits.py "<channel_url>" \
  --years 2 \
  --top 5 \
  --max-results 200

Parameters

Flag	Default	Description
channel_url	(required)	YouTube channel URL — accepts formats like @handle, /channel/ID, or full URLs
--years	2	How far back to look (supports decimals like 0.5 for 6 months)
--top	5	Number of top videos to return
--max-results	200	Max videos to pull from the API (increase for very active channels)
--json-output	off	Output machine-readable JSON instead of formatted text

The Apify token is not a command-line flag — it is read from the APIFY_API_KEY environment variable (see Prerequisites). This keeps tokens out of shell history.

Normalizing Channel URLs

The script handles most URL formats, but if the user gives you something unusual, normalize it before passing it in:

• https://www.youtube.com/@KevinStratvert/videos → https://www.youtube.com/@KevinStratvert
• @KevinStratvert → https://www.youtube.com/@KevinStratvert
• youtube.com/channel/UC4QZ_LsYcvcq7qOsOhpAI4A → pass as-is

The Script

#!/usr/bin/env python3
"""
Fetch a YouTube channel's top videos using the Apify YouTube Channel Scraper REST API.
Usage: python fetch_top_hits.py <channel_url> [--years 2] [--top 5]

Requires the APIFY_API_KEY environment variable (your own Apify token).
Uses only the requests library (no apify-client needed).
"""

import argparse
import json
import os
import sys
import time
import requests
from datetime import datetime, timedelta, timezone


ACTOR_ID = "streamers~youtube-scraper"
APIFY_BASE = "https://api.apify.com/v2"


def get_api_key():
    """Read the Apify token from the environment. Never fall back to a built-in key."""
    key = os.environ.get("APIFY_API_KEY", "").strip()
    if not key:
        sys.stderr.write(
            "ERROR: No Apify token found.\n\n"
            "This skill needs your own Apify API token. To get one:\n"
            "  1. Create a free account at https://apify.com\n"
            "  2. Copy your token from Settings -> API & Integrations\n"
            "  3. Set it as an environment variable, e.g.:\n"
            "       export APIFY_API_KEY=\"your_token_here\"   (macOS/Linux)\n"
            "       $env:APIFY_API_KEY=\"your_token_here\"      (PowerShell)\n\n"
            "Then re-run this command.\n"
        )
        sys.exit(1)
    return key


def run_actor(api_key, channel_url, max_results=200):
    """Start the Apify actor, poll until done, return dataset items."""
    url = f"{APIFY_BASE}/acts//runs"
    params = {: api_key}
    payload = {
        : [{: channel_url}],
        : max_results,
        : ,
        : ,
    }

    ()
    resp = requests.post(url, params=params, json=payload, timeout=)
     resp.status_code == :
        ()
        sys.exit()
     resp.status_code != :
        ()
        sys.exit()

    run_data = resp.json()[]
    run_id = run_data[]
    dataset_id = run_data[]
    ()

    
    status_url = 
    status = 
     i  ():  
        time.sleep()
        status_resp = requests.get(status_url, params={: api_key}, timeout=)
        status = status_resp.json()[][]
         status  (, , , ):
            
         i %  == :
            ()

     status != :
        ()
        sys.exit()

    ()

    
    items_url = 
    items_resp = requests.get(items_url, params={: api_key, : }, timeout=)
     items_resp.status_code != :
        ()
        sys.exit()

     items_resp.json()


 ():
    
     key  [, , , , ]:
        val = item.get(key)
         val   :
             (val, ):
                val = val.replace(, ).replace(, )
                :
                     (val)
                 ValueError:
                    
             (val)
    stats = item.get(, {})
     (stats, ):
        val = stats.get()  stats.get()
         val   :
             ((val).replace(, ))
     


 ():
    
     re
    m = re.search(, val.lower())
      m:
         
    qty = (m.group())
    unit = m.group()
    days_per = {: , : , : , : , :  / , :  / }
     datetime.now(timezone.utc) - timedelta(days=qty * days_per[unit])


 ():
    
     key  [, , , , , ]:
        val = item.get(key)
          val:
            
        sval = (val)
        :
               sval:
                 datetime.fromisoformat(sval.replace(, ))
             fmt  [, , , ]:
                :
                     datetime.strptime(sval, fmt).replace(tzinfo=timezone.utc)
                 ValueError:
                    
            rel = parse_relative_date(sval)
             rel   :
                 rel
         Exception:
            
     


 ():
     item.get()  item.get()  item.get()  


 ():
    url = item.get()  item.get()  item.get()
     url:
         url
    video_id = item.get()  item.get()
     video_id:
         
     


 ():
     key  [, , ]:
        val = item.get(key)
         val   :
             ((val).replace(, ))
    stats = item.get(, {})
     (stats, ):
        val = stats.get()  stats.get()
         val   :
             ((val).replace(, ))
     


 ():
     key  [, , , ]:
        val = item.get(key)
         val   :
             ((val).replace(, ))
     


 ():
     item.get()  item.get()  item.get()  


 ():
     n >= :
         
     n >= :
         
     (n)


 ():
    parser = argparse.ArgumentParser(description=)
    parser.add_argument(, =)
    parser.add_argument(, =, default=, =)
    parser.add_argument(, =, default=, =)
    parser.add_argument(, =, default=, =)
    parser.add_argument(, action=, =)
    args = parser.parse_args()

    api_key = get_api_key()

    
    channel_url = args.channel_url.rstrip()
     channel_url.endswith():
        channel_url = channel_url[: -()]

    ()
    ()

    start_time = time.time()
    dataset_items = run_actor(api_key, channel_url, args.max_results)
    elapsed = time.time() - start_time
    ()

      dataset_items:
        ()
        sys.exit()

    
    error_items = [it  it  dataset_items  (it, )  it.get()]
    dataset_items = [it  it  dataset_items  (it, )   it.get()]
     error_items:
        (
              )

      dataset_items:
        ()
        sys.exit()

    
    sample = dataset_items[]
    ()

    
    cutoff = datetime.now(timezone.utc) - timedelta(days=(args.years * ))
    filtered = []
    skipped_no_date = 

     item  dataset_items:
        dt = parse_date(item)
         dt  :
            skipped_no_date += 
            filtered.append(item)
         dt >= cutoff:
            filtered.append(item)

     skipped_no_date > :
        ()

    
    filtered.sort(key= x: parse_view_count(x), reverse=)
    top = filtered[: args.top]

     args.json_output:
        output = []
         i, item  (top, ):
            output.append({
                : i,
                : get_title(item),
                : get_url(item),
                : parse_view_count(item),
                : get_likes(item),
                : get_comments(item),
                : get_duration(item),
                : (parse_date(item)  ),
            })
        (json.dumps(output, indent=))
    :
        ()
        ()
        ()
         i, item  (top, ):
            views = parse_view_count(item)
            likes = get_likes(item)
            comments = get_comments(item)
            dt = parse_date(item)
            date_str = dt.strftime()  dt  

            ()
            ()
            ()
            ()
            ()

        total_views = (parse_view_count(v)  v  filtered)
        avg_views = total_views // (filtered)  filtered  
        ()
        ()
        ()


 __name__ == :
    main()

Presenting Results

After the script runs, present the results conversationally. Highlight what makes each video notable — is it a tutorial that clearly hit a nerve? A timely topic? Give the user context, not just a data dump. Mention the channel's overall stats (total views, average views per video) to put the top hits in perspective.

If the user wants to dig deeper (e.g., "what topics do best?", "show me a trend"), you can use the --json-output flag and do further analysis in Python.

Troubleshooting

• "No Apify token found" / 401 Unauthorized: The APIFY_API_KEY environment variable is missing or invalid. Follow the Prerequisites to set a valid token, then re-run.
• Actor fails or times out: The Apify free tier has limited compute. Suggest reducing --max-results to 50-100, or checking their Apify account balance.
• No videos returned: Double-check the channel URL is valid and public. Try the @handle format.
• View counts are 0: The API output field names may have changed. The script tries multiple field name variations, but if all views show as 0, print a sample item's raw JSON to debug: use --json-output and inspect the fields.
• Rate limits: Apify has rate limits on the free tier. If the user hits them, wait a minute and retry.