Quick Start - Automated Prompt Library Generation

TL;DR: Generate 90 prompt files across 15 categories in 3 sessions using automation + AI orchestration.

Created: 2025-11-29 Time Required: 5-8 hours total (across 3 sessions) Quality: 90+/100 (matching Aviation 98/100 exemplar)

⚡ 30-Second Quick Start

# Install dependencies
pip install pyyaml colorama tqdm

# Preview what will be generated
python3 generate-categories.py --priority high --dry-run

# Generate Session 1 (6 categories, 36 files)
python3 generate-categories.py --priority high

# Continue with AI orchestration for content
# (Use Task tool with orchestrator - see below)

📊 What You're Building

Final Output (After 3 Sessions)

prompts/
├── 01-educational-content/
│   ├── instructor-segments/           (6 files)
│   ├── concept-explanations/          (6 files)
│   ├── procedure-demonstrations/      (6 files)
│   └── assessment-content/            (6 files)
│
├── 02-technical-training/
│   ├── aviation/                      ✅ COMPLETE (exemplar)
│   ├── technology/                    (6 files)
│   └── professional/                  (6 files)
│
├── 03-b-roll-footage/
│   ├── establishing-shots/            (6 files)
│   ├── detail-shots/                  (6 files)
│   └── transition-shots/              (6 files)
│
├── 04-animations/
│   ├── diagrams/                      (6 files)
│   ├── processes/                     (6 files)
│   └── data-visualization/            (6 files)
│
└── 05-specialized/
    ├── multilingual/                  (6 files)
    ├── accessibility/                 (6 files)
    └── interactive/                   (6 files)

Total: 16 categories × 6 files = 96 files (90 new + 6 Aviation existing)

🎯 3-Session Roadmap

Session 1: High-Priority (2-3 hours)

Goal: Generate 6 most important categories (36 files)

Step 1: Generate Framework (5 minutes)

# This creates directory structure + README.md + CLAUDE.md
python3 generate-categories.py --priority high

# Output:
# ✅ instructor-segments: 6/6 files (86.2 KB)
# ✅ concept-explanations: 6/6 files (87.1 KB)
# ✅ procedure-demonstrations: 6/6 files (88.5 KB)
# ✅ technology: 6/6 files (85.9 KB)
# ✅ diagrams: 6/6 files (86.7 KB)
# ✅ accessibility: 6/6 files (85.0 KB)

What's Created:

✅ README.md (fully generated, user-facing guide)
✅ CLAUDE.md (fully generated, AI automation patterns)
⏸️ basic-template.md (placeholder - needs content)
⏸️ advanced-template.md (placeholder - needs content)
⏸️ examples.md (placeholder - needs content)
⏸️ customization-guide.md (placeholder - needs content)

Step 2: Generate Template Content (2-3 hours using AI orchestration)

Option A: One Category at a Time (Highest Quality - Recommended)

# For each of the 6 categories, use Claude Code Task tool:

In Claude Code conversation:

Use orchestrator subagent to complete instructor-segments category:

1. Read Aviation exemplar files:
   - 02-technical-training/aviation/basic-template.md
   - 02-technical-training/aviation/advanced-template.md
   - 02-technical-training/aviation/examples.md
   - 02-technical-training/aviation/customization-guide.md

2. Extract structural patterns and quality elements

3. Generate 4 files for instructor-segments:
   - basic-template.md (5 simple 10-20 word prompts for HeyGen instructor content)
   - advanced-template.md (5 production 150-300 word prompts with timing)
   - examples.md (10 tested prompts with results, costs, lessons learned)
   - customization-guide.md (adaptation guide for instructor content)

4. Use category config:
   Platform: HeyGen
   Cost: $0.016/sec
   Use cases: Course intros, concept explanations, demos, assessments

5. Quality target: 90+/100 (matching Aviation 98/100)

Location: .coditect-docs/docs/06-curriculum/video-generation-toolset/prompts/01-educational-content/instructor-segments/

Repeat for remaining 5 high-priority categories (20-30 min each).

Option B: Batch All 6 Categories (Faster, 85%+ Quality)

Use orchestrator subagent to batch-generate template content for 6 high-priority categories:

Categories:
1. instructor-segments (HeyGen, $0.016/sec) - Talking head presentations
2. concept-explanations (Veo 3.1, $0.15-0.40/sec) - Abstract visualizations
3. procedure-demonstrations (Runway Gen-3, $0.025/sec) - Step-by-step how-tos
4. technology (Veo 3.1, $0.15-0.40/sec) - Software tutorials
5. diagrams (Veo 3.1, $0.15-0.40/sec) - Flowcharts, architecture
6. accessibility (HeyGen, $0.016/sec) - WCAG compliance content

For EACH category:
1. Extract patterns from Aviation exemplar
2. Generate 4 template files (basic, advanced, examples, customization)
3. Adapt for category-specific use cases and platforms
4. Quality target: 85+/100

Token budget: 120K
Output: 24 files (6 categories × 4 files)

Location: .coditect-docs/docs/06-curriculum/video-generation-toolset/prompts/

Session 1 Output:

✅ 36 files total (6 categories × 6 files)
✅ 6 README.md + 6 CLAUDE.md (fully generated)
✅ 24 template files (AI-generated content)
✅ Quality: 90+/100 (Option A) or 85+/100 (Option B)

Session 2: Medium-Priority (2-3 hours)

Goal: Generate next 6 categories (36 files)

Step 1: Generate Framework (5 minutes)

python3 generate-categories.py --priority medium

Categories:

assessment-content
professional
establishing-shots
detail-shots
processes
data-visualization

Step 2: Generate Template Content (2-3 hours)

Use same approach as Session 1 (Option A or B).

Tip: Apply lessons learned from Session 1 for faster generation.

Session 2 Output:

✅ 36 files total (6 categories × 6 files)
✅ Cumulative: 72/90 files (80% complete)

Session 3: Low-Priority + Master Index (1-2 hours)

Goal: Complete remaining 3 categories + create navigation (18 files + index)

Step 1: Generate Framework (3 minutes)

python3 generate-categories.py --priority low

Categories:

transition-shots
multilingual
interactive

Step 2: Generate Template Content (45-60 minutes)

Use same approach as Session 1/2.

Step 3: Generate Master Index (15-30 minutes)

In Claude Code:

Create master index and navigation for complete prompt library:

1. Generate MASTER-INDEX.md with:
   - Complete category catalog (16 categories including Aviation)
   - Platform comparison matrix
   - Cost estimation guide
   - Use case cross-reference

2. Update parent README.md files:
   - prompts/README.md (add completion status)
   - Video generation toolset README.md
   - CODITECT docs README.md

3. Generate CATEGORY-REFERENCE.md with:
   - Related categories
   - Workflow combinations
   - Common multi-category use cases

Location: .coditect-docs/docs/06-curriculum/video-generation-toolset/prompts/

Session 3 Output:

✅ 18 files (3 categories × 6 files)
✅ Master index and navigation
✅ Cumulative: 90/90 files (100% complete)

✅ Quality Checklist

After each session, validate quality:

Per Category (6 files)

☐ README.md exists (8-10 KB, user-facing guide)
☐ CLAUDE.md exists (15-18 KB, AI automation)
☐ basic-template.md has 5 simple prompts (10-20 words each)
☐ advanced-template.md has 5 production prompts (150-300 words each)
☐ examples.md has 10 tested prompts with results/costs/lessons
☐ customization-guide.md has adaptation guide (18-22 KB)

☐ All files reference correct primary platform
☐ All cost estimates match platform pricing
☐ All use cases specific to category
☐ All compliance notes appropriate
☐ No broken cross-references

☐ Quality score >= 85/100 (minimum acceptable)
☐ Quality score >= 90/100 (target)

Quality Scoring

Score	Status	Action
90-100	EXCELLENT	Production-ready, publish
85-89	GOOD	Minor improvements optional
75-84	ACCEPTABLE	Functional, plan enhancements
<75	NEEDS IMPROVEMENT	Regenerate with fixes

🔧 Troubleshooting

Issue: Script Won't Run

# Check Python version (need 3.7+)
python3 --version

# Install dependencies
pip install pyyaml colorama tqdm

# Verify config exists
ls category-config.yaml

Issue: Low Quality Scores (<85)

Cause: Not following Aviation exemplar patterns closely enough.

Fix:

# Review Aviation exemplar
cat 02-technical-training/aviation/basic-template.md

# Use Task tool for analysis
Use codebase-analyzer subagent to extract Aviation quality patterns:

Analyze what makes Aviation category 98/100 quality:
1. Structural elements
2. Content density
3. Prompt quality
4. Documentation completeness

Apply findings to regenerate low-scoring category.

Issue: Token Budget Exhausted

Cause: Generating too many categories in one Task call.

Fix:

# Break into smaller batches (3 categories instead of 6)
# OR do one category at a time (Option A)

📚 Key Files Reference

Automation Framework

File	Purpose	Size
automation-framework.md	Complete framework documentation	30+ KB
category-config.yaml	Category definitions & metadata	15 KB
generate-categories.py	Automated generation script	20+ KB
quick-start-automation.md	This quick start guide	12 KB

Aviation Exemplar (98/100 Quality)

File	Purpose	Size
README.md	User-facing guide	9.3 KB
CLAUDE.md	AI automation	16.5 KB
basic-template.md	5 simple prompts	4.2 KB
advanced-template.md	5 production prompts	14.8 KB
examples.md	10 tested prompts	12+ KB
customization-guide.md	Adaptation guide	19.2 KB

🎯 Success Metrics

Framework Complete When:

✅ All 90 files generated (15 categories × 6 files)
✅ Average quality score >= 90/100
✅ Master index created
✅ All cross-references valid
✅ No empty placeholders remaining

Current Status:

✅ Aviation category: 100% complete (98/100 quality)
✅ Automation framework: Complete
⏸️ Remaining 15 categories: Ready for Session 1

💡 Pro Tips

Tip 1: Use Dry Run First

# Always preview before generating
python3 generate-categories.py --priority high --dry-run

Tip 2: Generate One Category First

# Test workflow with single category before batch
python3 generate-categories.py --category instructor-segments

Tip 3: Save Orchestrator Prompts

# Save successful prompts for reuse
cat > orchestrator-prompt-template.txt <<EOF
Use orchestrator subagent to complete {CATEGORY} category...
EOF

Tip 4: Track Quality Scores

# Keep log of quality scores per category
echo "instructor-segments: 92/100" >> quality-log.txt

Tip 5: Apply Lessons Learned

Session 1 teaches optimal prompting patterns
Session 2 goes faster with learned patterns
Session 3 is fastest (3 categories in 1 hour possible)

📞 Need Help?

Automation Issues:

Review automation-framework.md
Check category-config.yaml syntax
Verify dependencies installed

Quality Issues:

Study Aviation exemplar
Use codebase-analyzer to extract patterns
Increase specificity in orchestrator prompts

Content Issues:

Review platform documentation
Validate cost estimates
Check use case alignment

Time to Complete: 5-8 hours (across 3 sessions) Quality Target: 90+/100 Based on Exemplar: Aviation Category (98/100) Created: 2025-11-29

Part of CODITECT Training Framework - Video Generation Toolset

Ready to start? Run: python3 generate-categories.py --priority high --dry-run

⚡ 30-Second Quick Start​

📊 What You're Building​

Final Output (After 3 Sessions)​

🎯 3-Session Roadmap​

Session 1: High-Priority (2-3 hours)​

Session 2: Medium-Priority (2-3 hours)​

Session 3: Low-Priority + Master Index (1-2 hours)​

✅ Quality Checklist​

Per Category (6 files)​

Quality Scoring​

🔧 Troubleshooting​

Issue: Script Won't Run​

Issue: Low Quality Scores (<85)​

Issue: Token Budget Exhausted​

📚 Key Files Reference​

Automation Framework​

Aviation Exemplar (98/100 Quality)​

🎯 Success Metrics​

💡 Pro Tips​

Tip 1: Use Dry Run First​

Tip 2: Generate One Category First​

Tip 3: Save Orchestrator Prompts​

Tip 4: Track Quality Scores​

Tip 5: Apply Lessons Learned​

📞 Need Help?​