🌿 Alternef Digital Garden

Fabric AI

4 min read

Edit on GitHub

Fabric is an open-source framework for augmenting humans using AI, created by Daniel Miessler in January 2024. It provides a modular system for solving specific problems using a crowdsourced collection of AI prompts called “Patterns” that can be integrated into any workflow.

Philosophy

“AI isn’t a thing; it’s a magnifier of a thing. And that thing is human creativity.”

Fabric addresses the AI integration problem rather than the capabilities problem. While thousands of AI applications exist, integrating AI functionality into daily workflows remains challenging. Fabric solves this by:

  1. Breaking problems into components - Applying AI to individual pieces one at a time
  2. Organizing prompts by real-world tasks - Making it easy to discover, collect, and manage AI solutions
  3. Providing a unified interface - Command-line focused with REST API and web interface options

Key Features

Patterns System

Patterns are Fabric’s fundamental unit - carefully crafted prompts organized by task. The framework includes 240+ built-in patterns covering:

  • Content Analysis: extract_wisdom, summarize, analyze_claims, analyze_paper
  • Security: analyze_malware, analyze_threat_report, create_stride_threat_model, create_sigma_rules
  • Content Creation: create_academic_paper, write_latex, create_keynote, create_newsletter_entry
  • Development: create_coding_feature, create_design_document, analyze_logs, code_review
  • Visualization: create_mermaid_visualization, create_markmap_visualization, create_conceptmap
  • Personal: analyze_personality, create_reading_plan, create_flash_cards

Pattern Structure

Patterns use Markdown for maximum readability and editability:

# IDENTITY and PURPOSE
You extract surprising, insightful, and interesting information...
 
# STEPS
- Extract a summary of the content in 25 words...
- Extract 20 to 50 of the most surprising ideas...
 
# OUTPUT INSTRUCTIONS
- Only output Markdown.
- Write bullets as exactly 16 words.

Prompt Strategies

Fabric implements advanced prompting techniques:

  • Chain-of-Thought (cot): Step-by-step reasoning
  • Chain-of-Draft (cod): Iterative drafting with minimal notes
  • Tree-of-Thought (tot): Multiple reasoning paths with selection
  • Atom-of-Thought (aot): Break problems into atomic sub-problems
  • Least-to-Most (ltm): Solve from easiest to hardest
  • Self-Consistency: Multiple reasoning paths with consensus
  • Self-Refine: Answer, critique, and refine
  • Reflexion: Answer, brief critique, refined answer

Supported AI Providers

Fabric supports numerous AI providers:

  • OpenAI (GPT-4, o1, etc.) with Responses API
  • Anthropic (Claude Opus 4.5, Sonnet 4, etc.) with OAuth authentication
  • Google Gemini with TTS and web search
  • Ollama (local models)
  • Azure OpenAI
  • Amazon Bedrock
  • Perplexity AI with citations
  • Together AI
  • Venice AI (privacy-focused)
  • GitHub Models
  • Microsoft 365 Copilot (enterprise)
  • Digital Ocean GenAI
  • LM Studio, Groq, Mistral, and more

Installation

Unix/Linux/macOS:

curl -fsSL https://raw.githubusercontent.com/danielmiessler/fabric/main/scripts/installer/install.sh | bash

Windows PowerShell:

iwr -useb https://raw.githubusercontent.com/danielmiessler/fabric/main/scripts/installer/install.ps1 | iex

Package Managers

# macOS
brew install fabric-ai
 
# Arch Linux
yay -S fabric-ai
 
# Windows
winget install danielmiessler.Fabric
 
# From source (requires Go)
go install github.com/danielmiessler/fabric/cmd/fabric@latest

Docker

docker run --rm -it kayvan/fabric:latest --version

Usage Examples

Basic Pattern Usage

# Summarize clipboard content
pbpaste | fabric --pattern summarize
 
# Analyze claims with streaming
pbpaste | fabric --stream --pattern analyze_claims
 
# Extract wisdom from YouTube video
fabric -y "https://youtube.com/watch?v=VIDEO_ID" --stream --pattern extract_wisdom
 
# Scrape and analyze a website
fabric -u https://example.com -p analyze_claims

Advanced Features

# Use specific model
fabric -m gpt-4o --pattern summarize
 
# Apply prompt strategy
echo "Analyze this code" | fabric --strategy cot -p analyze_code
 
# Enable web search (supported models)
fabric --search --pattern research_topic
 
# Generate images
fabric --pattern create_art_prompt --image-file output.png
 
# Speech-to-text
fabric --transcribe-file audio.mp3 --pattern summarize
 
# Dry run (preview without API call)
echo "test" | fabric --dry-run -p summarize

REST API Server

# Start API server
fabric --serve
 
# With Ollama compatibility mode
fabric --serve --serveOllama

Custom Patterns

Create private patterns that won’t be overwritten during updates:

# Set up custom patterns directory
fabric --setup  # Select "Custom Patterns" option
 
# Create custom pattern
mkdir -p ~/my-custom-patterns/my-analyzer
echo "You are an expert analyzer..." > ~/my-custom-patterns/my-analyzer/system.md
 
# Use custom pattern
fabric --pattern my-analyzer "analyze this"

Custom patterns take precedence over built-in ones with the same name.

Helper Tools

  • to_pdf: Convert LaTeX to PDF, works with write_latex pattern
  • code2context: Generate JSON representation of code directories for AI analysis
  • generate_changelog: Generate changelogs from git history with optional AI summaries

Integration Options

  • Shell Aliases: Create direct commands for each pattern
  • Obsidian Integration: Save outputs directly to markdown vault
  • Shell Completions: Tab completion for Zsh, Bash, and Fish
  • Web Interface: Built-in GUI alternative to CLI
  • Per-Pattern Model Mapping: Configure specific models for individual patterns via environment variables

Technical Details

  • Language: Go (migrated from Python in 2024)
  • License: MIT
  • Platforms: Linux, macOS, Windows (including ARM)
  • Internationalization: Full i18n support for 10+ languages
  • API Documentation: Swagger/OpenAPI at /swagger/index.html

Resources

Backlinks

Graph View