Fabric 🧩 Workflow Open source

What it is

Fabric by Daniel Miessler (MIT) is a command-line framework built around "patterns": carefully written, crowdsourced system prompts for specific jobs such as extracting wisdom from a video transcript, summarising articles, writing essays, creating social posts, improving writing or analysing claims. You pipe text into a pattern and get structured output, which makes it handy for turning long content into notes, newsletters and posts. Because patterns are plain Markdown files in a public repo, you can read exactly what each prompt asks the model to do, copy them into ChatGPT or Claude without the CLI, and fork them into your own house style.

Key features

• Hundreds of Markdown prompt patterns you can read and edit
• Pipe-friendly CLI: feed in text, YouTube transcripts or web pages
• Works with OpenAI, Anthropic, Google Gemini, Ollama and other vendors
• Custom patterns folder for your own prompts
• REST API server and web interface options
• Very active development (Go rewrite)

How to install

Run the installer curl -fsSL https://raw.githubusercontent.com/danielmiessler/fabric/main/scripts/installer/install.sh | bash, or go install github.com/danielmiessler/fabric/cmd/fabric@latest, or use the Docker image, then fabric --setup to add API keys and download patterns.

Pricing and rights

Free and MIT-licensed, including the patterns. You pay for whichever model provider you connect; local models through Ollama cost nothing extra.

Who it is for

Writers, researchers and newsletter creators who work in a terminal and want repeatable summarising and drafting prompts.

Verdict

A well-designed, heavily used prompt library with a practical CLI. It is terminal-first, so non-technical users will prefer the prompts on their own over the tooling.