danielmiessler/fabric

Fabric is an open-source command-line framework for integrating AI into everyday tasks through a collection of reusable, composable prompts called "patterns." The project's core argument is that AI does not have a capabilities problem, it has an integration problem. There are plenty of powerful AI models available, but using them effectively in your daily workflow requires constructing the right prompts each time, which most people do not do consistently. Fabric addresses this by providing a crowdsourced library of carefully crafted prompts, each designed for a specific real-world task. Examples include summarizing content, extracting key insights from a YouTube video, writing security analyses, creating essay outlines, identifying cognitive biases in text, or generating social media posts. You pipe content into fabric on the command line, specify which pattern to apply, and receive a focused, high-quality response. Here is how it works: Fabric is a command-line tool written in Go. You install it, connect it to one or more AI providers (it supports OpenAI, Anthropic, Google Gemini, Ollama for local models, Azure OpenAI, and many others), and then run commands like echo "article text" | fabric --pattern summarize. Patterns are plain-text files containing system prompts stored in a local directory. You can use the built-in patterns, modify them, or write your own. Patterns can also be chained together so the output of one feeds into the next. You would use Fabric when you want AI capabilities woven into your terminal workflow, summarizing articles as you read them, processing transcripts, extracting action items from meeting notes, or applying consistent analysis to any piece of text without switching to a chat interface. The stack is Go for the CLI, with a REST API server option and a web UI for non-terminal users.