packaging-youtube-thumbnails

Generate, render, and validate YouTube title-thumbnail packages with performance learning.

3.6k|688|Updated Mar 28, 2026
Ready to run
npx skills add https://github.com/ericosiu/ai-marketing-skills --skill packaging-youtube-thumbnails-ericosiu
💬 Prompt for ChatGPT / Claude Chat▼
Please help me install this Agent Skill.
Skill: packaging-youtube-thumbnails
Source: https://github.com/ericosiu/ai-marketing-skills/tree/main/packaging-youtube-thumbnails
Command: npx skills add https://github.com/ericosiu/ai-marketing-skills --skill packaging-youtube-thumbnails-ericosiu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires Pillow, and includes scripts (resource) and references (resource) components.

What problem does it solve? Creating effective YouTube titles and thumbnails requires balancing channel identity, hook differentiation, brand accuracy, and mobile legibility, and most creators lack a repeatable, evidence-backed workflow for doing it. ## Core Features & Use Cases - Differentiated Package Generation: Creates materially different verdict, proof, and utility title-thumbnail packages scored against a weighted rubric with simplicity and semantic-clarity gates. - Deterministic QA Guards: Validates channel profiles, enforces numeric safe margins and timestamp-safe zones on 1280x720 renders, and detects drift outside allowed edit regions during surgical revisions. - Performance Learning Loop: Stores sanitized YouTube Studio readbacks in a local JSONL ledger, generates pre-package briefs, runs postmortem diagnoses, and promotes lessons only after three comparable 72-hour evidence videos. - Use Case: A creator supplies a transcript and channel profile, receives three scored packages with rendered variants shown inline, gives feedback, and gets a tested production handoff ZIP with finals, previews, and edit notes. ## Quick Start Use $packaging-youtube-thumbnails with my channel profile and this transcript to create three materially different title-thumbnail packages, score them, render the passing variants, and show the comparison set inline.

Frequently Asked Questions about packaging-youtube-thumbnails

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create YouTube title-thumbnail packages with this skill?▼

Supply a channel profile and source content such as a transcript or brief, then request a number of packages. The skill creates materially different verdict, proof, and utility packages, scores them against a rubric, and renders passing variants inline.

How do I validate a YouTube thumbnail before publishing?▼

Run thumbnail_guard.py with the final command, passing the image and safe-box coordinates for each critical brand mark. It checks 1280x720 dimensions, 64px horizontal and 36px vertical margins, and the 154x72 timestamp-safe corner.

Does this skill require a channel profile to work?▼

A usable channel profile is resolved from a user-supplied path or the current project. If none exists, the skill bootstraps one by auditing roughly 20 recent long-form thumbnails and saving the identity rules for reuse.

Can it learn from YouTube Studio analytics?▼

Yes, sanitized Studio readbacks are stored in a local performance ledger via thumbnail_learning.py. Postmortem diagnoses require at least three comparable records, and lessons are promoted only after three distinct 72-hour evidence videos with explicit review.

What are the limitations of the thumbnail editing workflow?▼

Surgical edits are validated by an outside-region diff guard, and any drift rejects the attempt. After one failed retry the result is labeled a controlled re-render rather than a surgical edit, and publishing always requires explicit approval.