mckinsey-slide
Create McKinsey-style slide decks as polished .pptx files: Pyramid Principle, MECE structure, SCR narrative, action titles, and pptxgenjs output.
Best use case
mckinsey-slide is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Create McKinsey-style slide decks as polished .pptx files: Pyramid Principle, MECE structure, SCR narrative, action titles, and pptxgenjs output.
Teams using mckinsey-slide should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/mckinsey-slide/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How mckinsey-slide Compares
| Feature / Agent | mckinsey-slide | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Create McKinsey-style slide decks as polished .pptx files: Pyramid Principle, MECE structure, SCR narrative, action titles, and pptxgenjs output.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
<!-- AUTO-GENERATED from SKILL.md.tmpl — do not edit directly -->
<!-- Regenerate: bun run gen:skill-docs -->
## Update Check (run first)
```bash
_UPD=$(~/.claude/skills/lenshub/scripts/lenshub-update-check 2>/dev/null || true)
[ -n "$_UPD" ] && echo "$_UPD" || true
```
If output shows `UPGRADE_AVAILABLE <old> <new>`: read `~/.claude/skills/lenshub/skills/lenshub-upgrade/SKILL.md` and follow the "Inline upgrade flow" (auto-upgrade if configured, otherwise AskUserQuestion with 4 options, write snooze state if declined). If `JUST_UPGRADED <from> <to>`: tell user "Running lenshub v{to} (just updated!)" and continue.
> "Start with the conclusion. The reader should never have to wonder where you are going."
> — Barbara Minto, *The Pyramid Principle*
# 📊 MCKINSEY SLIDE CREATOR
## Philosophy
You are not here to make pretty slides. You are here to **build a logical argument** — then render it visually. McKinsey's method: the conclusion comes first (Pyramid Principle), the structure is airtight (MECE), and every slide title is an insight not a topic. If you covered the body text and read only the titles, the story must still be complete.
**Core frameworks applied in order:**
1. **Pyramid Principle** (Barbara Minto): Lead with the governing thought. Support with 2-4 MECE arguments. Back each argument with evidence.
2. **SCR Arc**: Situation (what's true now) → Complication (why that's a problem) → Resolution (what we recommend).
3. **Action Titles**: Every headline is an insight sentence ("Revenue declined 23% due to customer churn in SMB") not a topic label ("Revenue").
4. **So What? Test**: Every slide must answer the "so what?" — if it doesn't, rewrite or cut it.
5. **Cover Test**: Hide all body text. Read only titles. The full story must survive.
6. **Rule of 3-5**: Max 3-5 bullets per slide. Max 7 words per bullet. One key message per slide.
Do NOT make the deck until you have completed Steps 0–2. Structure first, content second.
## Step 0: Context Intake
AskUserQuestion to collect (all required before proceeding):
1. **Topic / problem statement** — what decision or situation does this deck address?
2. **Audience** — board, C-suite, external client, internal team, investor?
3. **Key message** — complete this sentence: "After this deck, the audience will believe _____."
4. **Available data / evidence** — what supporting facts, numbers, or analysis exist?
5. **Deck length** — executive brief (3-5 slides), pitch (8-12 slides), full analysis (15-20 slides)?
6. **Output format** — structured markdown only, or generate pptxgenjs code for .pptx?
**STOP.** AskUserQuestion. Do NOT proceed until user provides context.
## Step 1: Pyramid Structure
Build the logical argument tree before writing a single slide.
```
GOVERNING THOUGHT (= Executive Summary headline)
├── ARGUMENT 1 (MECE bucket 1)
│ ├── Evidence 1a
│ └── Evidence 1b
├── ARGUMENT 2 (MECE bucket 2)
│ ├── Evidence 2a
│ └── Evidence 2b
└── ARGUMENT 3 (MECE bucket 3)
├── Evidence 3a
└── Evidence 3b
```
**MECE check on arguments:** Do they overlap? (If yes — merge or split.) Do they cover all the ground? (If gaps — add or acknowledge.) Mark: ✅ MECE or ❌ MECE VIOLATION.
Present the pyramid to the user. AskUserQuestion to confirm or adjust before proceeding.
## Step 2: SCR Narrative Arc
Map the deck sections to the SCR arc:
```
SITUATION | What is currently true? (shared context, no controversy)
COMPLICATION | Why is that a problem or opportunity? (the tension)
RESOLUTION | What should we do? (the recommendation)
```
If the key message is a recommendation: full SCR arc.
If the key message is a finding/analysis: Situation + Complication, Resolution = "therefore" summary.
**STOP.** AskUserQuestion to confirm the SCR framing. Do NOT proceed until confirmed.
## Step 3: Deck Architecture
Generate the full slide outline based on the pyramid and SCR arc:
```
SLIDE # | SECTION | ACTION TITLE (draft) | EVIDENCE
---------|---------------|-----------------------------------------|----------
1 | Cover | [Topic + audience + date] | —
2 | Exec Summary | [Governing thought as insight] | "So What?" test ✅
3 | Situation | [Action title] | [data]
4 | Situation | [Action title] | [data]
5 | Complication | [Action title] | [data]
...
N-1 | Recommendation| [Action title] | [data]
N | Next Steps | [3 concrete actions with owners/dates] | —
Appendix | Appendix | [Supporting detail] | —
```
## Step 4: Framework Pass
For each slide in the approved deck outline, read `references/frameworks.md` for full descriptions, visual layouts, and markdown rendering specs. Select the appropriate McKinsey framework and annotate each slide. No content yet.
Read `references/visual-identity.md` before writing any slide content.
Framework types: Pyramid Principle | MECE Issue Tree | McKinsey 7S | Three Horizons | GE-McKinsey Matrix | Value Bridge | Structured Action Plan
## Step 5: Action Title Pass
For every slide in the deck outline, apply this test to the title:
**So What? Test:** Does this title tell the audience what to conclude?
- ❌ Topic title: "Market Share" → ✅ Insight title: "We lost 8 points of market share in 18 months, concentrated in Gen Z"
- ❌ Topic title: "Customer Feedback" → ✅ Insight title: "NPS dropped 22 points because onboarding takes 3x too long"
Rewrite any title that fails. Every title must be a complete declarative sentence.
**Cover Test:** Read only the action titles in sequence. Does the story make sense without any body content? If not, restructure.
## Step 6: Content Generation
Write each slide in full:
---
### Slide [N]: [ACTION TITLE]
**Bullets** (3-5 max, 7 words max each, MECE):
- [Bullet 1]
- [Bullet 2]
- [Bullet 3]
**Visual Proof Point:** [Chart type + data description]
- Suggested chart: [bar / waterfall / scatter / table / 2x2 matrix / process flow]
- Key data: [what the visual shows]
**Speaker Notes:** [1-2 sentences the presenter says aloud — context not on slide]
---
Repeat for every slide in the deck.
If output format is pptxgenjs: after all slides are written, generate a complete `slides.ts` file using the pptxgenjs code patterns from `references/visual-identity.md`.
## Output
Produce the complete deck as structured markdown. Sections:
1. Pyramid structure (reference)
2. SCR arc (reference)
3. Full slide-by-slide content (the deliverable)
4. pptxgenjs `slides.ts` (if output format = pptx)
If the user wants to save: Write to `slides.md` (or `slides.ts`) in the current directory.
## CRITICAL RULES
- Every slide title MUST be an insight (declarative sentence), NEVER a topic label
- Every deck MUST pass the Cover Test: read titles only — story complete without body
- MECE check is mandatory on every group of bullets: no overlap, no gaps
- The So What? test applies to every slide — if a slide can't answer it, cut or merge it
- Maximum 3-5 bullets per slide; maximum 7 words per bullet
- One key message per slide — if you're making two points, you have two slides
## CRITICAL RULE — How to ask questions
Every AskUserQuestion MUST: (1) present 2-3 concrete lettered options, (2) state which you recommend FIRST, (3) explain WHY in 1-2 sentences. No batching. No yes/no questions.
## Completion Summary
```
+====================================================================+
| 📊 MCKINSEY SLIDE CREATOR — COMPLETION SUMMARY |
+====================================================================+
| Topic | [topic] |
| Audience | [audience] |
| Slides generated | ___ slides + ___ appendix slides |
| Pyramid structure | ___ arguments, MECE ✅ / ❌ |
| SCR arc | Situation / Complication / Resolution |
| Cover test | PASS / FAIL |
| Action titles | ___ of ___ passed So What? test |
| Output format | Markdown / pptxgenjs (.pptx) |
| Output | Markdown deck / Written to slides.md|ts |
+====================================================================+
"The reader should never have to wonder where you are going."
```Related Skills
ey-slide
Create EY Consulting-style slide decks as polished .pptx files. Uses the POINT narrative, 5-Box executive summary, and EY visual identity.
bcg-slide
Create BCG-style strategic slide decks as polished .pptx files: hypothesis-driven approach, issue tree, ghost deck workflow, SCQ narrative, BCG frameworks, and pptxgenjs output.
plan-musk
Elon Musk's first-principles engineering review. Deletes requirements, challenges every constraint, demands 10x thinking. "The best part is no part."
plan-jobs
Steve Jobs's product review. Obsesses over simplicity and user experience. Says no to 1000 things. Starts from what the user feels, not the architecture.
plan-feynman
Richard Feynman's clarity and rigor review. If you can't explain it simply, you don't understand it. Surfaces complexity debt, unjustified assumptions, and cargo-cult engineering. Best for: plans that feel complex but you can't articulate why.
plan-board
Board-level meta-synthesis. Runs Musk, Jobs, and Altman lenses in sequence, surfaces where they agree (GREEN) and disagree (RED), and delivers a final weighted verdict.
plan-bezos
Jeff Bezos's customer-obsession review. Works backwards from the press release. Asks: is this for the customer or for us? Will this matter in 10 years? Best for: plans where you're not sure if you're solving the right problem.
plan-altman
Sam Altman's review. Compounding leverage, AI-native thinking, moat building. "Does this matter in 10 years? Is there a version 10x better because of AI?"
lenshub-upgrade
Upgrade lenshub to the latest version. Pulls from the remote repo, regenerates all SKILL.md files, and shows what's new.
browse-qa
Post-plan-review smoke test. Uses the browser to verify the feature you planned actually shipped as intended. Produces a paste-ready QA report for PR descriptions and GitHub issues.
frontend-slides
Create stunning, animation-rich HTML presentations from scratch or by converting PowerPoint files. Use when the user wants to build a presentation, convert a PPT/PPTX to web, or create slides for a talk/pitch. Helps non-designers discover their aesthetic through visual exploration rather than abstract choices.
google-slides-automation
Lightweight Google Slides integration with standalone OAuth authentication. No MCP server required. Full read/write access.