skills/create-marp-deck/SKILL.md
Creates polished Marp slide decks through an interview-driven workflow. Use when the user wants to create a presentation, build slides, or make a deck. Interviews the user first (goal, audience, key points, data, constraints), then generates structured Marp Markdown with gradient section dividers, breadcrumb navigation, consistent formatting, and automatic HTML/PPTX export.
npx skillsauth add omerr/claude-skills create-marp-deckInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Create a presentation slide deck using the Marp markdown format, following the established conventions below.
Before generating anything, interview the user by asking these questions one at a time, waiting for each answer before proceeding to the next.
Question 1 — Goal & Topic: Ask: "What's this presentation about? What should the audience walk away knowing or being able to do?"
Question 2 — Audience: Ask: "Who's the audience? (e.g., engineering team, executives, external clients, conference attendees)"
Question 3 — Key Points: Ask: "What are the key points or sections you want to cover? List the main ideas, even if rough."
Question 4 — Data & Demos: Ask: "Any specific data, code examples, diagrams, or demos to include? Do you have these ready, or should I create them?"
Question 5 — Constraints: Ask: "Any constraints? Think: target number of slides, time limit for the talk, things to avoid, or specific branding requirements."
After collecting all answers, briefly summarize what you understood and ask for confirmation before proceeding.
Follow the structure and conventions below exactly.
Every deck starts with this exact frontmatter:
---
marp: true
theme: default
paginate: true
size: 16:9
---
Immediately after the frontmatter, include a <style> block with:
linear-gradient(135deg, #1e3a5f 0%, #2d5a8e 100%)linear-gradient(135deg, #064e3b 0%, #047857 100%)linear-gradient(135deg, #7a4a1a 0%, #a66b2e 100%)linear-gradient(135deg, #3d1e5c 0%, #5a2d8e 100%)linear-gradient(135deg, #0f172a 0%, #1e293b 50%, #0f172a 100%)<style>
section {
font-size: 22px;
}
table {
font-size: 18px;
}
blockquote {
font-size: 22px;
}
pre {
font-size: 16px;
}
header {
font-size: 14px;
color: #999;
}
header strong {
color: #2563eb;
}
section.title-slide {
background: linear-gradient(135deg, #0f172a 0%, #1e293b 50%, #0f172a 100%);
}
section.part-FIRST,
section.part-SECOND,
section.part-THIRD,
section.title-slide {
--h1-color: #fff;
--heading-strong-color: #fff;
--fgColor-default: rgba(255, 255, 255, 0.95);
--fgColor-muted: rgba(255, 255, 255, 0.7);
color: white;
}
section.part-FIRST {
background: linear-gradient(135deg, #1e3a5f 0%, #2d5a8e 100%);
}
section.part-SECOND {
background: linear-gradient(135deg, #064e3b 0%, #047857 100%);
}
section.part-THIRD {
background: linear-gradient(135deg, #3d1e5c 0%, #5a2d8e 100%);
}
</style>
Replace FIRST, SECOND, THIRD with short descriptive names for each part (e.g., part-context, part-results, part-summary).
The deck follows this exact order:
<!-- _class: lead title-slide -->
# Main Title
## Subtitle
**Key detail 1**: value
**Key detail 2**: value
**Date**: Month Year
List 3-5 questions the deck will answer, as a numbered list with bold lead-ins.
<!-- _class: lead title-slide -->
# Agenda
### Part 1: Name
Short description
### Part 2: Name
Short description
### Part 3: Name
Short description
Section divider (full-bleed colored background):
<!-- _header: "" -->
<!-- _class: lead part-NAME -->
# Part N: Section title
**One-line description**
Content slides (with breadcrumb header):
<!-- header: "**Active section** > Other section > Other section" -->
# Slide title
Content here
The breadcrumb shows where you are in the deck. The active section is wrapped in **bold**. The header directive persists until changed, so you only need to set it once per section (on the first content slide after the divider). Reset it with <!-- _header: "" --> before each section divider.
End with summary/takeaway slides, still within the last section.
-) for parenthetical separators, not double dash (--)>) for key callouts or distinctionsAfter writing the deck, export to both HTML and PPTX:
npx @marp-team/marp-cli@latest --no-stdin FILENAME.md -o FILENAME.html
npx @marp-team/marp-cli@latest --no-stdin --pptx FILENAME.md -o FILENAME.pptx
Always use --no-stdin to prevent marp from hanging on stdin.
If the user asks for editable PPTX (text that can be edited in PowerPoint or Google Slides), run these additional steps. This requires LibreOffice and python-pptx to be installed.
# PPTX - editable text (experimental, requires LibreOffice)
npx @marp-team/marp-cli@latest --no-stdin --pptx --pptx-editable --allow-local-files FILENAME.md -o FILENAME-editable.pptx
python3 -c "
from pptx import Presentation
from pptx.util import Emu
prs = Presentation('FILENAME-editable.pptx')
margin = Emu(747720)
for slide in prs.slides:
for shape in slide.shapes:
if shape.has_text_frame and shape.shape_type == 17:
tf = shape.text_frame
if not tf.text.strip() or tf.text.strip().isdigit():
continue
font_size = None
if tf.paragraphs and tf.paragraphs[0].runs:
font_size = tf.paragraphs[0].runs[0].font.size
if not font_size:
continue
new_width = prs.slide_width - margin - shape.left
if new_width > shape.width:
shape.width = new_width
min_height = int(font_size * 1.4) * 2
if shape.height < min_height:
shape.height = min_height
tf.word_wrap = True
prs.save('FILENAME-editable.pptx')
"
Notes:
--allow-local-files needed for local images--pptx-editable requires LibreOffice installed--no-stdin to prevent marp from hanging on stdindevelopment
Maintainer-only workflow for handling GitHub Secret Scanning alerts on OpenClaw. Use when Codex needs to triage, redact, clean up, and resolve secret leakage found in issue comments, issue bodies, PR comments, or other GitHub content.
development
Maintainer workflow for OpenClaw releases, prereleases, changelog release notes, and publish validation. Use when Codex needs to prepare or verify stable or beta release steps, align version naming, assemble release notes, check release auth requirements, or validate publish-time commands and artifacts.
development
Run, watch, debug, and extend OpenClaw QA testing with qa-lab and qa-channel. Use when Codex needs to execute the repo-backed QA suite, inspect live QA artifacts, debug failing scenarios, add new QA scenarios, or explain the OpenClaw QA workflow. Prefer the live OpenAI lane with regular openai/gpt-5.4 in fast mode; do not use gpt-5.4-pro or gpt-5.4-mini unless the user explicitly overrides that policy.
development
End-to-end Parallels smoke, upgrade, and rerun workflow for OpenClaw across macOS, Windows, and Linux guests. Use when Codex needs to run, rerun, debug, or interpret VM-based install, onboarding, gateway smoke tests, latest-release-to-main upgrade checks, fresh snapshot retests, or optional Discord roundtrip verification under Parallels.