Alex Banner Generation

Alex Banner Generation

Generate visually consistent SVG banners for any document in this heir using the Alex — ACT Edition brand template.

When to Use

• User asks for a banner, header image, or document decoration
• Creating a significant new document (README, PLAN-*.md, ROADMAP, CHANGELOG)
• User mentions adding a "header" or "branded image" to a doc

Looking for a lighter, hand-authored variant? The Mall ships document-banner-pastel -- pastel 1200x240 banners with content-specific iconography (tracks / hub-and-spokes / mockup / badge / symbol). Use that pattern for branding, education, or audience-facing docs; use this muscle for technical artifacts that need brand-stamped consistency.

Brand Constants (do not change)

Watermark Categories

Pick the one that matches the document's role:

If no category fits, ask the user before inventing one — the muscle rejects unknown watermarks.

Procedure

Step 1 — Gather inputs

Ask the user only for what's missing. Defaults:

• Title — the document's name. Keep ≤ 32 chars (the muscle enforces this).
• Subtitle — a single-line purpose statement, ≤ 80 chars. Lift it from the doc's first paragraph or its north-star sentence; don't invent.
• Watermark — pick from the table above based on doc role.
• Filename — defaults to assets/banner-<title-slug>.svg. Override with --out if the user wants a specific path.

Step 2 — Generate

node .github/muscles/generate-banner.cjs \
  --title "Document Title" \
  --subtitle "One-line purpose statement." \
  --watermark PLAN

Add --force to overwrite an existing file. Add --out path/to/banner.svg for a non-default location.

The muscle exits 0 on success, 1 on validation errors (length, watermark whitelist), 2 on filesystem errors (file exists without --force).

Step 3 — Embed in the document

Add this line just under the document's H1:

![Banner](assets/banner-<slug>.svg)

The muscle prints the exact embed line; copy it verbatim.

Subtitle Craft (the LLM-judgment part)

The muscle takes whatever subtitle you pass — quality is your job. Good subtitles:

• State the document's purpose, not its contents ("Critical thinking made operational." not "This document explains ACT.")
• Are one clause, not a sentence list
• End with a period
• Avoid hype ("revolutionary", "ultimate") and meta language ("this document")
• Match the document's actual first paragraph — don't promise things the doc doesn't deliver

If you're not sure the subtitle is right, show two options to the user before generating.

Validation Checklist

Before declaring done:

• File written under assets/
• Watermark matches the document's role (not just convenient)
• Title ≤ 32 chars, subtitle ≤ 80 chars (else the muscle rejects)
• Embed line added under the document's H1
• Renders in VS Code preview without errors

PNG Conversion (optional)

GitHub renders SVG banners natively in README.md and most surfaces, so SVG is preferred. If a downstream tool needs PNG:

# Via mermaid-cli's bundled chrome (already a dependency):
npx svgexport assets/banner-foo.svg assets/banner-foo.png 1200:300

Don't ship PNGs unless required — they double the asset weight and can drift from the SVG source.

Boundaries

• The muscle does not pick the watermark or write the subtitle — that is an LLM/skill judgment call.
• The muscle does not edit the source markdown — embedding the banner in the doc is a separate step.
• Custom colors / fonts / dimensions are not supported. If the user wants a non-template design, generate raw SVG manually rather than forking the muscle.

Falsifiability

• This skill is wrong if banners are consistently rejected aesthetically by the user, or if the SVG pipeline renders incorrectly in >10% of target environments (GitHub, VS Code preview, browsers)
• The template inventory is stale if the muscle adds new templates this skill does not document
• The color/font constraints are arbitrary if users frequently need customization the skill forbids