Marketing Skill

Skill: Parallel Tab Batch Processing for YouTube Comments

Overview

This skill enables you to open, read, and act on 10+ YouTube videos simultaneously instead of one at a time — dramatically cutting execution time. The core idea: launch all tabs in parallel, then process them in parallel batches.

Phase 1: Search & Extract Video IDs

Goal: Get a list of video IDs from a YouTube search results page.

// Run this in javascript_tool on the search results tab
const videos = document.querySelectorAll('a#video-title');
const results = [...videos].slice(0, 15).map(v => ({
  title: v.title,
  id: v.href?.split('v=')[1]?.split('&')[0]
})).filter(v => v.id && !v.id.startsWith('http')); // filter out Shorts/playlists

// Output as readable list
results.map(v => `${v.title} --> ${v.id}`).join('\n');

Tips:

• Grab 15 IDs to have backups in case some videos are unavailable
• Filter out Shorts (check that v.href contains /watch?v=, not /shorts/)
• Store the full list before opening any tabs

Phase 2: Open All Tabs in Parallel

Goal: Create N tabs simultaneously using browser_batch.

Step 1 — Create all tabs at once:
  browser_batch:
    - tabs_create (×10 — one per video slot)

Step 2 — Navigate all tabs in one batch:
  browser_batch:
    - navigate tab1 → youtube.com/watch?v=ID_1
    - navigate tab2 → youtube.com/watch?v=ID_2
    - navigate tab3 → youtube.com/watch?v=ID_3
    ... (all 10 in the same batch call)

Why this works: browser_batch executes sequentially per batch, but by launching all navigations together you avoid the serial wait of "open one → wait → open next."

Phase 3: Read All Tabs in Parallel

Goal: Extract descriptions and check for prior comments on all tabs at once.

Step 1 — Read all page text simultaneously:
  browser_batch:
    - get_page_text(tab1)
    - get_page_text(tab2)
    ...
    - get_page_text(tab10)

Step 2 — Scroll all tabs to load comments section:
  browser_batch:
    - scroll(tab1, direction=down, amount=15)
    - scroll(tab2, direction=down, amount=15)
    ...
    - scroll(tab10, direction=down, amount=15)

What to look for in the page text:

• @YOUR_CHANNEL_HANDLE → skip this video (already commented)
• Video title & description → use to craft your tailored comment
• "Video unavailable" → skip and use a backup ID from your list

Phase 4: Craft Tailored Comments

Goal: Write one unique comment per video based on its specific description.

Comment Formula:

[1 custom sentence about THIS video's specific topic or key insight]. [Your base promotional/informational text here.]

Tips for the custom sentence:

• Pull a specific detail from the description (a number, a date, a product name, a key claim)
• Connect it naturally to your base message's theme
• Keep it to 1–2 sentences max — don't over-write

Example:

Video topic: Company X announces 45 new satellite launches for 2025
Custom sentence: "Announcing 45 new satellite launches while also securing government contracts shows Company X is scaling fast — and understanding the broader market context matters more than ever."

Phase 5: Confirm with User (Required)

Goal: Present ALL proposed comments to the user before posting anything.

Format your proposal like this:

Here are the 10 comments I'd like to post:

**1. [Video Title]** (`video_id`)
> [Full proposed comment text]

**2. [Video Title]** (`video_id`)
> [Full proposed comment text]

...

Reply "yes" to post all comments.

Always wait for an explicit "yes" or confirmation before proceeding to Phase 6.

Phase 6: Post Comments in Batch

Goal: Submit all comments efficiently across all tabs.

For each tab (process 3–4 at a time to avoid rate limits):

Step A — Click the comment input box:
  find(tab, "Add a comment...")  → get element ref
  computer.left_click(ref)

Step B — Type the comment:
  computer.type(full_comment_text)

Step C — Submit:
  find(tab, "Comment submit button") → get element ref
  computer.left_click(ref)

Step D — Verify:
  computer.screenshot(tab)
  → confirm "@YOUR_CHANNEL_HANDLE just now" or "0 seconds ago" appears

Rate limit handling: If YouTube shows a posting error or rate limit, wait 10–15 seconds and retry. Process 3–4 tabs, pause briefly, then continue the next group.

Phase 7: Verification & Report

Goal: Confirm all comments are live and report to user.

✅ Posted: [Video Title] — @YOUR_CHANNEL_HANDLE confirmed (0 seconds ago)
✅ Posted: [Video Title] — @YOUR_CHANNEL_HANDLE confirmed (0 seconds ago)
...

Total: 10/10 comments posted successfully.

Decision Flowchart

Search YouTube for topic query
          ↓
Extract 15 video IDs via JavaScript
          ↓
Open 10 tabs in parallel (browser_batch: tabs_create × 10)
          ↓
Navigate all tabs in parallel (browser_batch: navigate × 10)
          ↓
Read all descriptions in parallel (browser_batch: get_page_text × 10)
          ↓
Scroll all tabs to load comments (browser_batch: scroll × 10)
          ↓
For each video:
  ├─ "Video unavailable" → skip, use backup ID from list
  ├─ "@YOUR_CHANNEL_HANDLE" found in comments → skip (already posted)
  └─ OK → craft tailored comment using description
          ↓
Present ALL proposed comments to user → WAIT for "yes"
          ↓
Post comments in groups of 3–4 (pause between groups)
          ↓
Screenshot-verify each → Report full completion

Key Efficiency Principles

Common Errors & Fixes

Customization Guide

This skill works for any YouTube topic and any comment template. To adapt it:

1. Change the search query → swap search_query= parameter in the YouTube URL
2. Change the comment base text → update your fixed promotional/informational message
3. Change the channel handle → replace @YOUR_CHANNEL_HANDLE with your actual handle
4. Change the number of tabs → adjust from 10 to any number (5–15 is practical)
5. Change the skip logic → update the duplicate-check string to match your channel handle

Notes

• This skill was developed and validated through live YouTube commenting sessions across 24 videos in 3 topic categories (SpaceX IPO, ASTS, Rocket Lab).
• Always confirm with the user before posting — never auto-post without explicit approval.
• YouTube's comment rate limits vary; if you hit errors, space out your submissions.
• This pattern generalizes beyond YouTube to any multi-tab read-then-act workflow.