Fabric: AI Prompt Patterns for Everyday Use

Meet Fabric

#

Fabric is an open-source tool by Daniel Miessler that provides 230+ reusable AI prompt patterns for common tasks like summarizing content, extracting insights, improving writing, and more.

Think of it as a library of expertly-crafted prompts you can apply to any content.

Why Fabric?

#

Most people interact with AI through free-form chat. That works, but it’s inconsistent. The same question asked differently produces different results.

Fabric solves this with patterns—structured prompts that produce reliable, formatted output every time. Instead of asking “can you summarize this?”, you run the summarize pattern and get consistent results.

Installation

#

Prerequisites

#

• Go 1.21+ (install via brew install go on macOS)

Install Fabric

#

go install github.com/danielmiessler/fabric/cmd/fabric@latest

Add Go’s bin directory to your PATH (add to ~/.zshrc or ~/.bashrc):

export PATH="$PATH:$HOME/go/bin"

Initial Setup

#

Create the config directory and download patterns:

mkdir -p ~/.config/fabric
touch ~/.config/fabric/.env
fabric -U

This downloads 230+ patterns to ~/.config/fabric/patterns/.

(Optional) Add API Keys

#

If you want Fabric to call AI APIs directly, run:

fabric -S

This walks you through configuring providers like OpenAI, Anthropic, Ollama, etc.

Basic Usage

#

List Available Patterns

#

fabric -l

Run a Pattern

#

Pipe content into Fabric:

cat article.txt | fabric -p summarize

Or use it with other tools:

curl -s https://example.com/article | fabric -p extract_wisdom
pbpaste | fabric -p improve_writing | pbcopy

YouTube Transcripts

#

Fabric can fetch and process YouTube videos:

fabric -y "https://youtube.com/watch?v=VIDEO_ID" -p extract_wisdom

Essential Patterns

#

Here are the patterns I use most often:

Pattern Deep Dive: extract_wisdom

#

This is Fabric’s signature pattern. It extracts:

• SUMMARY — 25-word overview
• IDEAS — Key concepts (16 words each)
• INSIGHTS — Refined, abstracted takeaways
• QUOTES — Notable quotes from the content
• HABITS — Practical habits mentioned
• FACTS — Verifiable facts stated
• RECOMMENDATIONS — Actionable advice

It’s perfect for processing podcasts, articles, videos, or books.

Native Pattern Execution

#

You don’t need Fabric’s CLI to use the patterns. Each pattern is just a markdown file with structured instructions.

Pattern location:

~/.config/fabric/patterns/{pattern_name}/system.md

You can read these files directly and use them with any AI tool—Claude, ChatGPT, or a local model. Just paste the pattern as a system prompt, then provide your content.

This is how I use Fabric patterns with Claude Code: I read the pattern file, apply it to content, and get structured output without spawning the CLI.

Creating Custom Patterns

#

Patterns follow a simple structure:

# IDENTITY and PURPOSE

You are an expert at [domain]. Your task is to [purpose].

# STEPS

1. [First step]
2. [Second step]
...

# OUTPUT INSTRUCTIONS

- Output in [format]
- Include [sections]

# INPUT

INPUT:

Save your pattern to:

~/.config/fabric/patterns/my_pattern/system.md

Then use it like any other:

cat content.txt | fabric -p my_pattern

Using Fabric Patterns in Claude Code

#

If you’re using Claude Code (or any AI assistant with file access), you can use Fabric patterns without the CLI. Just ask your assistant to apply a pattern to your content.

Example Prompts

#

Basic pattern application:

Use the extract_wisdom pattern on this: [paste content]

Apply the summarize pattern to this article: [paste or URL]

Run improve_writing on this text: [paste]

With files:

Use extract_wisdom on the transcript in ~/Downloads/podcast.txt

Apply analyze_paper to the PDF at ~/Documents/research.pdf

Web content:

Fetch https://example.com/article and run summarize on it

Output to file:

Use extract_wisdom on this and save the output to ~/notes/insights.md

What Happens Behind the Scenes

#

When you ask me to use a Fabric pattern, I:

1. Read the pattern from ~/.config/fabric/patterns/{pattern}/system.md
2. Apply those instructions to your content
3. Return structured output matching the pattern’s format

No API keys needed—I’m already the AI processing the content.

Quick Reference

#

You can also ask me to list patterns (what Fabric patterns are available?) or read a specific pattern to see what it does.

Tips

#

1. Chain patterns — Pipe output from one pattern into another
2. Use with clipboard — pbpaste | fabric -p pattern | pbcopy (macOS)
3. Update regularly — Run fabric -U to get new patterns
4. Read the patterns — Understanding how they work helps you use them better
5. Start with summarize and extract_wisdom — They handle 80% of use cases

Resources

#

• Fabric GitHub — Source code and documentation
• Pattern Directory — Browse all patterns online
• Daniel Miessler’s Blog — Creator’s writing on AI and security

Fabric turns AI from a chatbot into a reliable tool. Once you build the habit of reaching for patterns, you’ll wonder how you worked without them.