Fabric Patterns

Fabric Patterns - Intelligent Pattern Selection

When to Activate This Skill

• User mentions "fabric" or "fabric pattern"
• "What fabric pattern should I use for..."
• "Help me analyze/summarize/extract..."
• Processing content (articles, videos, podcasts, code)
• Creating structured outputs (summaries, reports, analyses)
• Transforming text between formats
• Security analysis, threat modeling, code review
• Any text-based workflow that could benefit from AI processing

What is Fabric?

Fabric is an open-source framework for augmenting human capabilities with AI. It organizes 242+ specialized AI prompts (called "Patterns") into task-based categories, making it easy to apply the right AI solution for any workflow.

Common usage: fabric --pattern <pattern_name>

Core Workflow: Pattern Selection

When user asks for help with Fabric:

1. Understand the Intent

Ask yourself:

• What is the user's primary goal? (extract, analyze, summarize, create, transform)
• What type of content? (text, code, video transcript, data)
• What output format? (summary, report, visualization, structured data)
• Is this a single step or multi-step workflow?

2. Match to Categories

EXTRACT - Pull specific information from content

• extract_wisdom - Insights, ideas, quotes, habits from any content
• extract_article_wisdom - Wisdom from articles specifically
• extract_ideas - Novel concepts and applications
• extract_recommendations - Actionable advice and guidance
• extract_questions - Q&A extraction
• extract_references - Citations and sources

SUMMARIZE - Condense content at various levels

• summarize - General-purpose summarization
• create_micro_summary - Concise one-sentence + key points
• create_5_sentence_summary - Progressive detail levels
• youtube_summary - YouTube videos with timestamps
• summarize_paper - Academic papers with findings

ANALYSIS - Evaluate, assess, compare

• analyze_claims - Truth evaluation with evidence
• analyze_paper - Scientific paper analysis
• analyze_prose - Writing quality evaluation
• compare_and_contrast - Side-by-side comparison tables
• find_logical_fallacies - Argument validity check
• rate_content - Multi-dimensional quality scoring

DEVELOPMENT - Code and software workflows

• explain_code - Code explanation and analysis
• review_code - Comprehensive code review
• create_git_diff_commit - Smart commit messages
• summarize_git_changes - Change summaries
• create_coding_project - Project design with architecture
• write_pull-request - PR descriptions

SECURITY - Cybersecurity and threat analysis

• create_stride_threat_model - STRIDE threat modeling
• analyze_malware - Malware behavior analysis
• analyze_threat_report - Threat intelligence extraction
• create_sigma_rules - Detection rule creation
• write_hackerone_report - Vulnerability reports

WRITING - Content creation and improvement

• improve_writing - Enhance clarity and style
• clean_text - Fix formatting and structure
• tweet - Convert to social media posts
• create_formal_email - Professional email composition
• write_essay - Essay creation in specific styles

VISUALIZE - Create diagrams and visual content

• create_mermaid_visualization - Mermaid diagrams
• create_markmap_visualization - Mind maps
• create_conceptmap - Interactive concept maps
• create_visualization - ASCII art with explanations

AI & PATTERNS - Meta-patterns for working with AI

• suggest_pattern - Get pattern recommendations from Fabric itself
• create_pattern - Design new custom patterns
• improve_prompt - Enhance AI prompts

3. Provide Output

Format your response with:

🎯 Recommended Patterns

Primary Pattern: pattern_name

• Why: [Explanation of why this pattern fits]
• Command:

fabric --pattern pattern_name

• With input file:

fabric --pattern pattern_name < input.txt

• From clipboard (macOS):

pbpaste | fabric --pattern pattern_name

• From URL:

fabric -u https://example.com/article --pattern pattern_name

• From YouTube:

fabric -y "https://youtube.com/watch?v=VIDEO_ID" --pattern pattern_name

Alternative Patterns:

• alt_pattern_1 - [When to use this instead]
• alt_pattern_2 - [Another approach]

🔗 Multi-Pattern Workflows

For complex tasks, suggest chaining patterns:

# Example: Deep content analysis pipeline
fabric --pattern extract_wisdom < article.txt > step1_wisdom.md
fabric --pattern capture_thinkers_work < step1_wisdom.md > step2_concepts.md
fabric --pattern label_and_rate < step2_concepts.md > final_categorized.md

📋 Ready-to-Use Scripts

Point to provided scripts in scripts/ directory:

• fabric_run_pattern.sh - Run single pattern on file/directory
• fabric_pipeline.sh - Chain multiple patterns together
• Custom scripts for common workflows

4. Pattern Categories Reference

Quick category lookup (20+ categories):

• AI: ai, create_pattern, improve_prompt, suggest_pattern
• ANALYSIS: analyze_claims, analyze_paper, compare_and_contrast
• BUSINESS: create_prd, check_agreement, create_loe_document
• CONVERSION: convert_to_markdown, export_data_as_csv, translate
• DEVELOPMENT: explain_code, review_code, create_git_diff_commit
• EXTRACT: extract_wisdom, extract_ideas, extract_recommendations
• SECURITY: create_stride_threat_model, analyze_malware, analyze_threat_report
• SUMMARIZE: summarize, create_micro_summary, youtube_summary
• WRITING: improve_writing, clean_text, create_formal_email
• VISUALIZE: create_mermaid_visualization, create_markmap_visualization

Full pattern database available in skill frontmatter and reference files.

Common Usage Patterns

Basic Usage

# List all available patterns
fabric --list

# Update patterns to latest version
fabric --updatepatterns

# Stream results in real-time
fabric --stream --pattern pattern_name

# Save output to file
fabric --pattern pattern_name < input.txt > output.md

# Use with specific model
fabric --model gpt-4 --pattern pattern_name

Advanced Usage

# With context file for additional background
fabric --pattern pattern_name --context background.md < input.txt

# Multiple inputs from directory
cat articles/*.md | fabric --pattern extract_wisdom > all_wisdom.md

# Chain with other tools
curl -s https://api.example.com/data | jq '.content' | fabric --pattern analyze_claims

Custom Patterns

Users can create custom patterns in ~/.config/fabric/patterns/pattern-name/system.md

Suggest using create_pattern pattern to design new ones:

echo "I need a pattern that does X" | fabric --pattern create_pattern

Scripts Provided

fabric_run_pattern.sh

Run any pattern on a file or directory:

./scripts/fabric_run_pattern.sh extract_wisdom ./articles/
./scripts/fabric_run_pattern.sh summarize ./transcript.txt output.md

fabric_pipeline.sh

Chain patterns in sequence:

# Default pipeline: extract_wisdom → capture_thinkers_work → label_and_rate
./scripts/fabric_pipeline.sh ./articles/

# Edit script to customize pipeline patterns

Key Commands Quick Reference

# Pattern selection help
fabric --pattern suggest_pattern

# List patterns
fabric -l

# Update patterns
fabric -u

# Stream output
fabric -s --pattern <name>

# From clipboard
pbpaste | fabric -p <name>

# From URL
fabric -u <url> -p <name>

# From YouTube
fabric -y <url> -p <name>

# Save output
fabric -p <name> -o output.md

Best Practices

1. Start with suggest_pattern - Let Fabric recommend patterns:

   echo "I want to extract key insights from research papers" | fabric --pattern suggest_pattern
   

2. Use streaming for long content - Get results faster:

   fabric -s --pattern extract_wisdom
   

3. Chain patterns for deep analysis - Each pattern refines the previous output

4. Create custom patterns - For repeated specialized tasks

5. Keep patterns updated - Run fabric --updatepatterns regularly

Examples

Example 1: Analyze Blog Post

# Copy blog post to clipboard, then:
pbpaste | fabric --stream --pattern analyze_claims

Example 2: Extract Wisdom from YouTube

fabric -y "https://youtube.com/watch?v=dQw4w9WgXcQ" --pattern extract_wisdom

Example 3: Code Review

git diff main | fabric --pattern review_code

Example 4: Create Threat Model

fabric --pattern create_stride_threat_model < system_design.md

Example 5: Multi-Step Research Analysis

# Pipeline approach
fabric -u https://research-paper-url.com --pattern summarize_paper > summary.md
fabric --pattern extract_wisdom < summary.md > wisdom.md
fabric --pattern create_markmap_visualization < wisdom.md > mindmap.md

Troubleshooting

Pattern not found: Run fabric --updatepatterns

No output: Check if fabric is configured with fabric --setup

Model errors: Verify API keys are set in ~/.config/fabric/.env

Want different model: Use --model flag or change default with fabric --changeDefaultModel

Reference Files

This skill includes:

• Complete pattern database with descriptions (references/pattern-database.md)
• Example scripts (scripts/)
• Fabric official documentation (references/fabric-readme.md)

Pattern Database

For the complete categorized list of all 242+ patterns with descriptions, see: $HOME/.claude/skills/productivity/fabric-patterns/references/pattern-database.md

Next Steps After Pattern Selection

1. ✅ Run the suggested command
2. 📄 Review the output
3. 🔄 Chain with another pattern if needed
4. 💾 Save results to appropriate location
5. 🎨 Customize pattern or create new one if necessary