fabric-ai
Guides pattern selection through decomposition, exploration, and diverse approach generation
When & Why to Use This Skill
The Fabric Orchestrator is a powerful Claude skill designed to optimize AI workflows by guiding precise pattern selection through deep decomposition and creative exploration. It leverages an extensive library of over 200 specialized patterns and multiple reasoning strategies to generate 5-7+ diverse approaches for any given challenge, ensuring high-quality, multi-dimensional solutions for complex tasks.
Use Cases
- Complex Problem Solving: Decomposing intricate challenges into multiple reasoned approaches, allowing users to choose the most effective strategy based on trade-offs and specific goals.
- Automated Content Extraction: Streamlining the processing of YouTube videos, web URLs, and audio files by applying specialized patterns like 'extract_wisdom' to generate structured insights.
- AI Workflow Orchestration: Managing multi-step AI sessions and pipelines by combining specific patterns and strategies to ensure consistency and depth in outputs across writing, analysis, and development tasks.
- Pattern Discovery and Selection: Navigating a vast library of 220+ patterns across various domains (Security, Business, Development) to find the most relevant prompt engineering technique for a specific use case.
| name | fabric-ai |
|---|---|
| description | Guides pattern selection through decomposition, exploration, and diverse approach generation |
Fabric Orchestrator
--ultrathink --seq
Core Mission
Explore patterns creatively → Generate 5-7+ deeply reasoned approaches per challenge
Resources: references/pattern_descriptions.json (228 patterns), references/strategies.json (9 strategies)
Workflow
- Analyze (--ultrathink --seq): Components, goals, dimensions, creative angles
- Ask (AskUserQuestion): Objective, constraints, creativity level, expansion ideas
- Consult: Read JSONs
- Generate: 5-7+ approaches (unique angle, patterns, strategy, WHY, commands, trade-offs)
- Present → Execute → Iterate
Pattern Discovery
- Read Resoureces
- Explore MULTIPLE tag categories (ANALYSIS:97, WRITING:63, DEVELOPMENT:44, EXTRACT:39, BUSINESS:34, SECURITY:30, others)
- Consider unexpected combinations
- Think across dimensions
Content Types
| Type | Flags |
|---|---|
| YouTube | -y <url> --transcript |
| Web | -u <url> |
| Audio/Video | --transcribe-file <path> |
Always add: --stream (long), --session (multi-step), -c (clipboard), -o (save), --strategy
Pipelines
fabric <input> -p <pattern> --session <name>
fabric -p <pattern> --strategy <strat> --session <name> -c
Quality Checklist
- Read both JSONs
- Explored MULTIPLE tags
- Generated 5-7+ approaches
- Deep WHY reasoning
- Varied all dimensions
- Used --session for pipelines
Commands
fabric -y <url> -p extract_wisdom --strategy cot --stream
fabric --listpatterns / --liststrategies / --listmodels