Skip to content
Back to skills

Analyse Channel Packaging

ASecurity

Extract a channel's profile and banner branding before generating new packaging. Triggered by "analyse my banner", "what is this channel's visual brand", "review my channel packaging". In Spanish, "analiza mi banner", "cómo es la marca visual de este canal", "revisa mi packaging de canal".

  • 2 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added October 7, 2026
ai-agentspythongoapi

Works with

  • api

Security analysis

A100/100

Scanned October 7, 2026

npx -y skills add lorena-bordonaba-pau/youtube-agent --skill analyse-channel-packaging --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Analyse Channel Packaging?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Analyse Channel Packaging
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/lorena-bordonaba-pau-analyse-channel-packaging/badge)](https://www.skillsdirectory.com/skills/lorena-bordonaba-pau-analyse-channel-packaging)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: analyse-channel-packaging
description: Extract a channel's profile and banner branding before generating new packaging. Triggered by "analyse my banner", "what is this channel's visual brand", "review my channel packaging". In Spanish, "analiza mi banner", "cómo es la marca visual de este canal", "revisa mi packaging de canal".
---

# Channel packaging analysis

## WHEN

Before generating a banner or a profile picture, your own or inspired by
another channel. Without seeing the current brand, whatever gets generated will
be generic or will break an identity that already works.

## EXECUTION ORDER

1. **Download and measure both pieces**
   ```
   python3 tools/view_channel_packaging.py [--channel ID] --md
   ```
   Returns the avatar and banner downloaded, with the resolution, contrast and
   saturation of each.

2. **Open both files** with the image reading tool. The pixels tell you whether
   there is contrast; they do not tell you what it conveys.

3. **Read the banner by its centre band.** This is the most common reading
   error: on mobile the top and bottom thirds are cropped away entirely. A
   banner is judged by what survives in that strip, not by what you see on a
   desktop. See `youtube-banner-spec`.

4. **Analyse the elements**
   - Colour scheme, and whether avatar and banner agree with each other.
   - Banner typography: tagline, hierarchy, legibility when small.
   - Avatar: is it recognisable at 32x32, its size in the comments?
   - Banner zones: what sits in the safe band and what is lost to the crop.
   - Identity elements: logos, signature colours, repeated motifs.

5. **Cross against the positioning.** Read `memory/channel_positioning.md`:
   the packaging either backs the channel's promise or contradicts it. That
   crossing is the analysis; describing colours is not.

6. **If you are going to generate**, load `youtube-banner-spec` or
   `youtube-profile-spec` depending on the piece, and pass the current
   packaging as a reference:
   ```
   python3 tools/generate_image.py --type banner --prompt "..." \
     --ref data/packaging/<channel_id>_banner.jpg:packaging --dry-run
   ```
   Identity elements are preserved unless a full, explicit rebrand was asked
   for.

## SOURCES

`view_channel_packaging` is `derived`: pixel metrics over images from the API.
The brand reading is the agent's judgement over files it has opened, and is
declared as such. `channel_positioning.md` is `config`, with its date.

## OUTPUT

A brief in 5-8 bullets: what works, what is lost on mobile, what contradicts
the positioning, and what would stay untouched in a redesign.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.