deck-assembly
Assembles a new HTML presentation by selecting content from existing PowerPoint source decks and rendering it in a polished Copilot-themed design.
Skill by zrosenfield
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/zrosenfield-deck-assembly
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: presentations, design, frontend, writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/deck-assembly
What's inside
- Skill name
deck-assembly- Description
- Assembles a new HTML presentation by selecting content from existing PowerPoint source decks and rendering it in a polished Copilot-themed design. Use when the user wants to build a new deck from content in other decks — triggers include "make a new deck using slides from," "combine these decks," "take the roadmap from this deck and the intro from that one," or "assemble a presentation for this customer." Produces a self-contained HTML file that opens in any browser, navigates with arrow keys, and looks like a premium dark-themed Microsoft AI briefing deck.
- License
- MIT
- Instructions
- 158 lines in SKILL.md · view on GitHub
Files (3)
- SKILL.md
- known_limitations.md · references/
- copilot-theme.html · templates/
SKILL.md (first lines)
# Deck Assembly
This skill reads content from PowerPoint decks and generates a complete HTML presentation. The visual output is always the Copilot dark theme — you never try to replicate the source deck's visual style. Instead you extract the content (numbers, quotes, key facts, speaker info) and choose the right visual treatment for each piece of information.
Do not generate Python scripts. Your output is the complete HTML file, delivered as text.
---
## The Core Principle
**Every piece of content has a natural visual form. Your job is to find it.**
A number is not text — it is a stat card or a statement slide. A quote is not a bullet — it is a quote slide. A name and title is not a paragraph — it is a profile slide. Before you decide to use bullets, ask whether the content is actually better shown as a number, a card, a bar, or a single focused visual.
When in doubt: **fewer words, bigger visual.**
---
## Content Pattern Recognition
Read these rules before making any slide type decisions. Apply them strictly.
### Numbers and percentages — always visual, never textual
If source content contains a number, percentage, dollar figure, ratio, or ranked score — regardless of how it appears in the source — it must become a visual element. It is never a bullet point.
| What you see in the source | What you use |
|---|---|
| One dominant number + 1–2 lines of context | `slide-statement` — number fills the slide |
| Two numbers being compared | `slide-stat cols-2` — one card each |
| Three or four numbers compared | `slide-stat cols-3` or `cols-4` |
| Percentages with category labels (survey, breakdown) | `slide-two-col` with `.bar-metric` rows |
| A before/after pair | `slide-stat cols-2` with contrasting card styles |
### Slide density signals the layout
Count the meaningful content items on the source slide — not counting the title.
| Content density | Layout to use |
|---|---|
| Title only, or ≤ 3 words total | `slide-section` or `slide-logo` |
| 1 number + 1–2 lines | `slide-statement` |
| 2–4 numbers or metrics | `slide-stat` |
| 1 person's name + role | `slide-profile` |
| 1 quoted sentence + attribution | `slide-quote` |
| 3–6 **named** capabilities, features, pillars, or roadmap items | `feature-grid` slide — **never bullets** |
| 3–5 parallel observations with no individual names | `slide-content` with bullet dots |
| 2 related but distinct blocks of content | `slide-two-col` |
| A photograph or image reference | `slide-image` |
| A brand mark or logo | `slide-logo` |
**The critical distinction:** If each item has its own name ("Copilot Pages", "Agent Marketplace", "Agentic Identity"), it is a named feature and must be a `feature-card`. If the items are observations or takeaways without individual names, bullets are acceptable.
### The bullet point is the last resort
Only use `slide-content` with bullets when the content is genuinely a list of parallel observations that have no individual names and cannot be better shown as numbers, cards, or a quote.
Ask yourself these questions in order before using bullets:
1. Does this slide have a headline number? → `slide-stat` or `slide-statement`
2. Does each item have its own name/title? → `feature-grid` with `feature-card` items
3. Is one item a quote? → `slide-quote`
4. Are there exactly two themes? → `slide-two-col`
Only if all four answers are "no" should you use bullets.
A slide that says "34% experts, 81% with AI, 87% AI alone" → three stat cards, not bullets.
A slide listing "Copilot Pages, Copilot Actions, Agent Marketplace, SharePoint+" → feature cards with icons, not bullets.
### Text that looks short is probably visual
PowerPoint speakers routinely put only 3–8 words on a slide because those words are meant to be the visual anchor for what they're saying, not a transcription. A slide with just "The Frontier Firm" is a `slide-section`. A slide with just "45%" and a caption sentence is a `slide-statement`. Do not pad these out into bullets.
---
## Slide Type Reference
Read `templates/copilot-theme.html`. That file defines every component, its CSS class names, and example HTML for each slide type. When generating the output, copy the HTML structure from that file exactly — do not improvise class names or invent new patterns.
**slide-title** — Deck opener. Badge pill + two-line H1 + subtitle. One or two orbs. Use for the first slide and dramatic section openers only.
**slide-stat** — One to four metric cards. Each card: large number + short label. Use `cols-1` through `cols-4`. Make one card `.hi` (highlighted, darker background, accent border) to indicate the most important or the "after" state. Use `grad-text` on the number you want to emphasize most; white on the others.
**slide-statement** — Single giant number dominates the entire slide. No card. Use for the one number that is the whole point of a slide. 160–180px, `grad-text`, short description below.
**slide-section** — Section break. Eyebrow label + accent bar + large left-aligned H2. Minimal. Use between major sections of the deck.
**slide-content** — Title + three to five bullet items with gradient dot markers. Use `<strong>` for the key phrase in each bullet. Keep bullets short — under 12 words each. This is the fallback only.
**slide-quote** — Giant gradient quotation mark, large centered quote text, attribution. Use for any attributed statement you find in the source.
**slide-profile** — Circular avatar ring + name + role + company. Use for speaker intros and customer spotlights. Use initials in the avatar when no photo is available.
**slide-two-col** — Title + two-column layout. Left column: bullets or bar metrics. Right column: a stat card, a second set of metrics, or supporting content. Use when you have two related but distinct ideas.
**feature-grid slide** — A plain `<div class="slide">` (no named slide class) with inline `display:flex;flex-direction:column;gap:28px` containing a title, optional subtitle, and a `<div class="feature-grid cols-N">`. Each item is a `<div class="feature-card [hi]">` with a `.feature-icon` (emoji), `.feature-name`, and `.feature-desc`. For roadmap slides, add a `.date-chip` above the icon. Use `cols-2` for 4 items (2×2), `cols-3` for 6 items (2×3), `cols-5` for a horizontal strip of 5 capabilities or milestones. Alternate `.hi` cards for visual rhythm.
**slide-logo** — Brand mark fills the slide. Extra-dark background. Use for company spotlights, logo reveals, or when a source slide has essentially nothing except a logo.
**slide-image** — Full-bleed image with dark overlay, text anchored to bottom. Use when the source references a photo, team shot, or the slide title alone suggests a visual moment ("The team", "Our journey", etc.). Use `class="img-bg placeholder"` when no actual image is available, and note this to the user.
---
## Visual Components Available Within Slides
These components can be composed inside any slide type. Use them to add visual richness without breaking the design system.
**`.bar-metric`** — A label, percentage value, and a horizontal fill bar. Use inside `col-card` or `slide-two-col` when content includes category-level percentages (e.g., talent breakdown, survey results by group). The bar width is set inline: `style="width: 75%"`.
**`.tag` / `.tag.accent`** — Small pill chip for category or status labels. Use inside stat cards to label "Before" vs "After" or category names.
**`.callout-num`** — A very large number (120px) used inline within a slide rather than inside a card. Useful when you want a number to dominate a content slide without a full card wrapper.
**`.divider`** — Thin horizontal separator. Use inside cards to separate two metrics stacked vertically.
**Gradient text** — Apply `class="grad-text"` to any text element to get the blue-to-lavender number gradient. Apply `class="brand-text"` for the purple-to-blue gradient. Use these on numbers, key words in titles, and the logo mark.
---From zrosenfield/sharepoint-ai-skills, MIT.
More from zrosenfield/sharepoint-ai-skills
- 0.639score
list-formatting
zrosenfield
Create beautiful, functional SharePoint list and library formatters using JSON.
SkillBFreshness: B★ 99Claude Code +9 - 0.639score
site-analyzer
Marcus Markiewicz
Performs comprehensive analysis of a SharePoint project site by systematically reading ALL lists and ALL documents using AI in SharePoint tools, then…
SkillBFreshness: B★ 99Claude Code +9 - 0.636score
autoresearch
Marcus Markiewicz
Optimize any AI in SharePoint skill by iteratively running it against test inputs, scoring outputs with binary evals, mutating the prompt to fix failures, and…
SkillBFreshness: B★ 99Claude Code +9 - 0.635score
improve-skills
zrosenfield
Sets up persistent eval suites for any AI in SharePoint skill and runs self-healing improvement loops on demand.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
faq-building
zrosenfield
Builds structured FAQ pages from source material such as documents, process guides, policies, product descriptions, or topic briefs.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
forest-style
zrosenfield
Applies the forest-style brand system to any visual output, document, web content, presentation, table, list, or interface element.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
copy-editing
zrosenfield
Systematically edits Markdown documents — READMEs, skill files, and demo guides — using seven sequential sweeps.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
executive-summary
zrosenfield
Distills long documents, reports, meeting transcripts, or briefing materials into tight one-page executive summaries formatted for SharePoint.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
hiring-pipeline-report
zrosenfield
Reads an Open Roles list and a Candidates list from the current SharePoint site and generates a self-contained HTML hiring dashboard for Zava Industries.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
ingest-knowledge
zrosenfield
Extracts tacit knowledge from any input — typed text, transcripts, emails, documents, or Slack threads — and writes it into the Knowledge Base library as…
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
kb-admin-report
zrosenfield
Generates a comprehensive knowledge base health report covering corpus overview, knowledge decay, gap status, ingestion activity, expert coverage, and…
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
gap-analysis
zrosenfield
Compares two documents or versions of content and surfaces what is missing, conflicting, changed, or new between them.
SkillBFreshness: B★ 99Claude Code +9
Signals
- Stars
- 99
- Forks
- 31
- Open issues
- 0
- Last commit (this folder)
- Aug 4, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from zrosenfield
- 0.639score
list-formatting
zrosenfield
Create beautiful, functional SharePoint list and library formatters using JSON.
SkillBFreshness: B★ 99Claude Code +9 - 0.635score
improve-skills
zrosenfield
Sets up persistent eval suites for any AI in SharePoint skill and runs self-healing improvement loops on demand.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
design-direction
zrosenfield
Creates a distinctive visual direction for interfaces, pages, dashboards, slides, forms, and other visual artifacts.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
faq-building
zrosenfield
Builds structured FAQ pages from source material such as documents, process guides, policies, product descriptions, or topic briefs.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
forest-style
zrosenfield
Applies the forest-style brand system to any visual output, document, web content, presentation, table, list, or interface element.
SkillBFreshness: B★ 99Claude Code +9 - 0.634score
executive-summary
zrosenfield
Distills long documents, reports, meeting transcripts, or briefing materials into tight one-page executive summaries formatted for SharePoint.
SkillBFreshness: B★ 99Claude Code +9
Cite this page
deck-assembly skill. Agent Mart. https://agentmart.dev/skills/zrosenfield-deck-assembly (last updated 2026-10-24)