TheMattBerman Slideshow Kit
Top-level router. Dispatches to subskills and workflows based on user intent.
When the user asks for a carousel
If they have a brand workspace at <brands_root>/<slug>/ (default ./brands/, see references/brand-management.md):
- List available styles with
scripts/list_styles.sh <slug>when the style is ambiguous or the user asks for a look that is not already defined. - If no suitable style exists, ask the user to describe the reusable style and
whether they have visual examples. Then follow
references/style-system.mdto create<brand>/styles/<style-name>/DESIGN.mdand optional refs before rendering. - Brand-aligned, social-native, and thread/native text looks all route through
skills/styled-carousel/with the selected--style.
If they don't have a brand workspace yet:
- Read
workflows/onboard-brand.mdand walk the user through paste mode (default) or interview mode - The workflow runs
scripts/init_brand.sh <slug>itself, then extracts brand DNA from samples or 12 questions - After completion, all three DNA files plus reusable style(s) are present at
<brands_root>/<slug>/; a default character is also present when a social-native/candid-person style is selected.
When the user asks to run the daily loop
(v0.3.0+) Read workflows/daily-loop.md and walk the steps.
When the user asks about brands
- Create new:
scripts/init_brand.sh <slug> - List existing:
scripts/list_brands.sh - Switch default:
scripts/switch_brand.sh <slug> - Create or refine a visual style: follow the natural-language flow in
references/style-system.md; usescripts/add_style.shonly as the internal implementation detail. - Multi-brand model: see
references/brand-management.md
Brand DNA contract
Every brand has three files at <brands_root>/<slug>/:
brand-voice.md— how the brand talksbrand-perspective.md— what it believes (pillars, hot takes, ICP)visual-system.md— palette, typography, layout, output sizes
Subskills and workflows read these directly. Never duplicate or vendor brand state inside the kit.