Skill: Client Showcase Generator

Trigger

"Create client showcase" / "Build portfolio demo" / "Demo for {prospect}" / "Show PMM Brain capabilities"

Process

Step 1: Load context

• Read data/brain/pmm_brain.json — Brain summary for system proof
• Scan data/output/ — Best output examples across formats
• Reference Obsidian vault structure at ~/Documents/Knowledge-brain-Obsidian/Knowledge Brain/PMM Brain/

Step 2: Determine showcase type

Based on user request, generate one or more:

#### A. Pipeline Demo (Full workflow walkthrough) Walk through a real example end-to-end:

1. Input: Theme selection from Brain (show suggest-themes output)
2. Research: Evidence sourcing from 13 newsletters, 861 raw posts, 37 mental models
3. Generation: 3-angle draft production with format selection
4. Critique: Rubber Duck Escalator scoring with real scores
5. Distribution: Pillar → micro atomization across 4 formats
6. Output: Final content package (LinkedIn + Twitter + Newsletter + Carousel)

#### B. Case Study (2-3 completed content packages) For each package:

1. Theme: What was written about and why
2. Research brief: What evidence and models informed it
3. Critique scores: Real Escalator scores showing quality bar
4. Final outputs: The actual content across formats
5. Results: Engagement metrics if available (optional)

#### C. Template Package (System architecture — exportable) Document the system as a replicable blueprint:

1. CLAUDE.md pattern: How the orchestrator routes between skills
2. Skill structure: How each skill works (trigger → process → output)
3. Brain schema: How knowledge is structured (mental models, evidence bank, topic layers, newsletter insights)
4. Quality gate: How the Rubber Duck Escalator ensures output quality
5. Distribution engine: How pillar content atomizes into platform-native formats

Step 3: Generate showcase artifacts

• Write in professional-but-personal tone (Sourav's voice, not corporate)
• Include real numbers: 13 newsletter sources, 37 mental models, 69 evidence stats, 861 raw posts
• Show before/after examples where possible
• Highlight what makes this system different from generic AI content tools

Step 4: Present to user

Show generated artifacts with clear sections. User can approve, tweak, or request different emphasis.

Output Format

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
CLIENT SHOWCASE: [Type]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

[Showcase content organized by type]

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
SYSTEM STATS
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Newsletter sources: 13
Raw post notes: 861
Mental models: 37
Evidence bank stats: 69
Content formats: 4 (LinkedIn, Twitter/X, Newsletter, Carousel)
Quality gate: 5-phase Rubber Duck Escalator (threshold: 8+)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━