claude-presentation-master 7.4.0 → 8.0.0

package/README.md

@@ -29,17 +29,19 @@ Claude Presentation Master transforms your content (Markdown, text, JSON, YAML)
 ```
 Your Content (Markdown, text, etc.)
         ↓
-  Content Analysis (detects presentation type)
+  Content Analysis (auto-detects presentation type)
         ↓
-  Slide Generation (KB-driven structure)
+  Pattern Classification → Structural Slide Types
+        ↓
+  Slide Generation (KB-driven, expert methodology)
         ↓
   Image Assignment (local → Unsplash → placeholder)
         ↓
-  7-Dimension QA Scoring
+  Visual Quality Evaluation (Playwright-based)
         ↓
   Auto-Fix Issues (iterates until 95+)
         ↓
-  Output: HTML (Reveal.js) + PowerPoint
+  Output: HTML (Reveal.js) + PDF + PowerPoint
 ```
 
 ---
@@ -127,6 +129,29 @@ If you want AI-generated images, create them separately and put them in your `im
 
 ---
 
+## 🏗️ v8.0.0 Unified Architecture
+
+Version 8.0.0 introduces a **unified type system** that ensures consistent behavior across all presentation types:
+
+### Semantic → Structural Type Mapping
+
+The knowledge base uses **semantic types** (e.g., `problem`, `solution`, `traction`) that the engine automatically maps to **structural types** (e.g., `single-statement`, `metrics-grid`, `three-column`):
+
+```
+Content Pattern → Semantic Intent → Structural Slide Type
+     ↓                  ↓                   ↓
+  "3 bullets"    →   "key points"    →  "three-points"
+  "$2.4M ARR"    →   "traction"      →  "metrics-grid"
+  "vs competitor"→   "competition"   →  "comparison"
+```
+
+This architecture ensures:
+- **Any content works** — not just specific test cases
+- **Universal 95+/100 scores** — tested across keynotes, investor pitches, consulting decks
+- **Consistent output** — same input always produces same quality
+
+---
+
 ## 🧠 The Knowledge Base
 
 This isn't a template library. It's a **73,000+ token expert system** containing:
@@ -190,9 +215,20 @@ experts:
 
 ---
 
-## 📊 7-Dimension Quality Scoring
+## 📊 Visual Quality Evaluation (Playwright-Based)
+
+Every presentation is scored using **real browser rendering** via Playwright. The tool evaluates actual visual output, not just metadata:
+
+### 4-Dimension Visual Scoring (25 points each)
 
-Every presentation is scored across 7 dimensions. The tool **iterates until 95+**:
+| Dimension | What It Measures |
+|-----------|------------------|
+| **Narrative Flow** | Logical progression, strong opening, clear transitions, memorable close |
+| **Visual Consistency** | Typography uniformity, color palette adherence, spacing consistency |
+| **Content Quality** | Word limits, readability, key message clarity, bullet structure |
+| **Executive Readiness** | Would impress stakeholders, boardroom-ready, compelling, shareworthy |
+
+### 7-Dimension Content Scoring (Backup/Additional)
 
 | Dimension | Weight | What It Measures |
 |-----------|--------|------------------|
@@ -207,7 +243,7 @@ Every presentation is scored across 7 dimensions. The tool **iterates until 95+*
 ### Iterative Auto-Fix
 
 ```
-Generate → Score (85/100) → Auto-Fix Issues → Regenerate → Score (92/100) → Auto-Fix → Score (97/100) → ✅ Done
+Generate → Visual Score (85/100) → Auto-Fix Issues → Regenerate → Score (95/100) → ✅ Done
 ```
 
 The engine will iterate up to 5 times (configurable) to reach your target score.
@@ -219,14 +255,17 @@ The engine will iterate up to 5 times (configurable) to reach your target score.
 ### Option 1: npx (No Install — Just Run)
 
 ```bash
-# Generate a keynote presentation instantly
+# Generate a keynote presentation (HTML + PDF auto-generated)
 npx claude-presentation-master generate notes.md -m keynote -f html
 
 # Generate a consulting deck as PowerPoint
 npx claude-presentation-master generate strategy.md --type consulting_deck -f pptx
 
-# Both formats
+# All formats (HTML + PDF + PPTX)
 npx claude-presentation-master generate content.md -f html,pptx
+
+# With verbose output
+npx claude-presentation-master generate content.md -f html --verbose
 ```
 
 ### Option 2: Global Install (Short Command)
@@ -286,7 +325,7 @@ COMMANDS:
 OPTIONS:
   -o, --output <dir>      Output directory (default: ./output)
   -m, --mode <mode>       keynote (6-25 words) or business (40-80 words)
-  -f, --format <fmt>      html, pptx, or html,pptx
+  -f, --format <fmt>      html, pptx, or html,pptx (PDF auto-generated with html)
   --type <type>           Presentation type (auto-detected if omitted)
   -t, --theme <name>      Theme: default, consulting-classic, modern-tech, minimal
   --title <title>         Presentation title
@@ -295,6 +334,7 @@ OPTIONS:
   --max-iterations <num>  Max auto-fix iterations (default: 5)
   --no-iterative          Disable iterative QA (single pass)
   --skip-qa               Skip QA (NOT recommended)
+  --verbose               Show detailed progress output
 
 EXAMPLES:
   # TED-style keynote
@@ -354,12 +394,18 @@ fs.writeFileSync('pitch.pptx', result.outputs.pptx);
 - Keyboard shortcuts: `S` (speaker notes), `F` (fullscreen)
 - Embedded CSS with presentation-type theming
 
+### PDF (Auto-generated from HTML)
+
+- Print-ready format
+- Generated automatically with HTML
+- High-fidelity rendering via Playwright
+- Perfect for email attachments
+
 ### PowerPoint (PPTX)
 
 - Opens in Microsoft PowerPoint
 - Fully editable
 - Send to clients
-- Print to PDF
 - Includes IB specialty charts (football field, waterfall, comps)
 
 ---

package/assets/presentation-knowledge.yaml

@@ -649,6 +649,7 @@ experts:
       glance_test:
         description: "Audience should comprehend each slide in 3 seconds"
         word_limit: 25
+        time_limit: "3 seconds"
       star_moment:
         description: "Something They'll Always Remember"
         purpose: "Create one dramatic memorable moment"
@@ -6361,6 +6362,7 @@ presentation_types:
       sources_required: false
 
     slide_types_allowed:
+      # Semantic types (TED keynote specific)
       - "title_impact"
       - "single_statement"
       - "big_number"
@@ -6369,6 +6371,14 @@ presentation_types:
       - "three_points"
       - "call_to_action"
       - "star_moment"
+      # Structural types (for pattern-based slide selection)
+      - "three_column"
+      - "metrics_grid"
+      - "comparison"
+      - "timeline"
+      - "process"
+      - "data_insight"
+      - "cta"
 
     required_elements:
       - "Opening hook (first 10 seconds)"
@@ -6429,6 +6439,7 @@ presentation_types:
       sources_required: false  # Social proof instead
 
     slide_types_allowed:
+      # Semantic types (sales pitch specific)
       - "title_impact"
       - "problem_statement"
       - "solution_overview"
@@ -6439,6 +6450,17 @@ presentation_types:
       - "pricing"
       - "call_to_action"
       - "demo_screenshot"
+      # Structural types (for pattern-based slide selection)
+      - "three_points"
+      - "three_column"
+      - "metrics_grid"
+      - "timeline"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "data_insight"
+      - "quote"
+      - "cta"
 
     required_elements:
       - "Problem statement early"
@@ -6510,6 +6532,7 @@ presentation_types:
       callouts_required: true  # For charts
 
     slide_types_allowed:
+      # Semantic types (consulting deck specific)
       - "executive_summary_scr"
       - "data_insight"
       - "comparison"
@@ -6518,6 +6541,18 @@ presentation_types:
       - "recommendation"
       - "risks_mitigation"
       - "next_steps"
+      # Structural types (for pattern-based slide selection)
+      - "three_points"
+      - "three_column"
+      - "two_column"
+      - "metrics_grid"
+      - "big_number"
+      - "timeline"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "quote"
+      - "cta"
 
     required_elements:
       - "Executive summary first (answer first)"
@@ -6605,6 +6640,7 @@ presentation_types:
       dense_data_allowed: true  # Tables, complex charts OK
 
     slide_types_allowed:
+      # Semantic types (investment banking specific)
       - "executive_summary"
       - "situation_overview"
       - "credentials"
@@ -6617,6 +6653,20 @@ presentation_types:
       - "sources_uses"
       - "process_timeline"
       - "risk_factors"
+      # Structural types (for pattern-based slide selection)
+      - "three_points"
+      - "three_column"
+      - "two_column"
+      - "metrics_grid"
+      - "big_number"
+      - "comparison"
+      - "timeline"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "data_insight"
+      - "quote"
+      - "cta"
 
     required_elements:
       - "Bank credentials and deal experience"
@@ -6699,6 +6749,7 @@ presentation_types:
       sources_required: true  # For market size, traction
 
     slide_types_allowed:
+      # Semantic types (investor pitch specific)
       - "title_impact"
       - "problem"
       - "solution"
@@ -6710,6 +6761,18 @@ presentation_types:
       - "team"
       - "financials"
       - "ask"
+      # Structural types (for pattern-based slide selection)
+      - "three_points"
+      - "three_column"
+      - "metrics_grid"
+      - "big_number"
+      - "comparison"
+      - "timeline"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "data_insight"
+      - "quote"
 
     sequoia_structure:
       order:
@@ -6789,6 +6852,7 @@ presentation_types:
       sources_required: true
 
     slide_types_allowed:
+      # Semantic types (technical presentation specific)
       - "title"
       - "architecture_diagram"
       - "data_flow"
@@ -6798,6 +6862,19 @@ presentation_types:
       - "metrics"
       - "tradeoffs"
       - "next_steps"
+      # Structural types (for pattern-based slide selection)
+      - "three_points"
+      - "three_column"
+      - "two_column"
+      - "metrics_grid"
+      - "big_number"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "data_insight"
+      - "quote"
+      - "cta"
+      - "full_image"
 
     required_elements:
       - "Clear problem context"
@@ -6864,6 +6941,7 @@ presentation_types:
       sources_required: false  # Internal data OK unsourced
 
     slide_types_allowed:
+      # Semantic types (all hands specific)
       - "title_impact"
       - "metrics_grid"
       - "celebration"
@@ -6873,6 +6951,18 @@ presentation_types:
       - "milestones"
       - "three_points"
       - "call_to_action"
+      # Structural types (for pattern-based slide selection)
+      - "three_column"
+      - "two_column"
+      - "big_number"
+      - "comparison"
+      - "timeline"
+      - "process"
+      - "bullet_points"
+      - "single_statement"
+      - "data_insight"
+      - "quote"
+      - "cta"
 
     required_elements:
       - "Start with wins/celebrations"

package/bin/cli.js

@@ -32,6 +32,7 @@ USAGE:
 COMMANDS:
   generate <input>    Generate presentation from input file
   validate <file>     Validate an existing HTML presentation
+  visual-qa <file>    Run HIGH-LEVEL visual quality evaluation (Playwright)
   info                Show package information
 
 OPTIONS:
@@ -176,6 +177,7 @@ function parseArgs(args) {
 
       case 'generate':
       case 'validate':
+      case 'visual-qa':
       case 'info':
         options.command = arg;
         break;
@@ -314,6 +316,8 @@ Usage:
       await runGenerate(inputPath, options, generate);
     } else if (options.command === 'validate') {
       await runValidate(inputPath, options, validate);
+    } else if (options.command === 'visual-qa') {
+      await runVisualQA(inputPath, options);
     }
   } catch (error) {
     if (error.code === 'ERR_MODULE_NOT_FOUND') {
@@ -585,6 +589,53 @@ Category Breakdown:
   }
 }
 
+// Visual QA command - HIGH-LEVEL quality evaluation
+async function runVisualQA(inputPath, options) {
+  console.log(`
+╔════════════════════════════════════════════════════════╗
+║          Claude Presentation Master                    ║
+║          HIGH-LEVEL Visual Quality Evaluation          ║
+╚════════════════════════════════════════════════════════╝
+`);
+
+  console.log(`📄 Evaluating: ${inputPath}`);
+  console.log(`📸 Taking screenshots with Playwright...`);
+  console.log('');
+
+  try {
+    const { VisualQualityEvaluator } = await import('../dist/index.js');
+
+    const outputDir = resolve(process.cwd(), options.output);
+    const screenshotDir = join(outputDir, 'qa-screenshots');
+
+    const evaluator = new VisualQualityEvaluator(screenshotDir);
+    const result = await evaluator.evaluate(inputPath);
+
+    // Display the report
+    console.log(evaluator.generateReport(result));
+
+    // Exit with appropriate code
+    process.exit(result.overallScore >= (options.threshold || 80) ? 0 : 1);
+
+  } catch (error) {
+    if (error.message?.includes('Playwright')) {
+      console.error(`
+❌ Playwright Error
+
+Make sure Playwright is installed:
+  npx playwright install chromium
+
+Error details: ${error.message}
+`);
+      process.exit(1);
+    }
+
+    console.error('Error:', error.message);
+    console.error(error.stack);
+    process.exit(1);
+  }
+}
+
 // Run
 main().catch(error => {
   console.error('Fatal error:', error);