danielmiessler/Fabric

What it solves

Fabric solves the "integration problem" of AI by organizing AI prompts into a structured framework. Instead of managing thousands of disparate chat-bots and apps, it allows users to create, collect, and organize fundamental AI solutions—called "Patterns"—in one place and integrate them directly into their existing workflows, particularly via the command line.

How it works

Fabric treats prompts as the primary unit of AI capability. It provides a CLI tool and a REST API server that can be connected to a wide variety of AI providers (including OpenAI, Anthropic, Google Gemini, Ollama, and others). Users can apply specific "Patterns" (pre-defined prompts for real-world tasks) to their input data. It also supports per-pattern model mapping, allowing users to assign specific AI models to specific tasks.

Who it’s for

It is designed for people who want to integrate AI into their daily lives and work, especially those who are command-line focused or developers who want to programmatic access to AI prompts via an API.

Highlights

• Extensive Provider Support: Integrates with dozens of native and OpenAI-compatible AI vendors.
• Pattern-Based Workflow: Organizes prompts by real-world tasks like summarizing academic papers, extracting YouTube highlights, or explaining code.
• CLI-First Interface: Optimized for terminal use with support for shell aliases to run patterns as direct commands.
• Flexible Deployment: Available as a binary, via package managers (Homebrew, Winget, Scoop), or as a Docker container.
• REST API: Includes a server with Swagger/OpenAPI documentation for easier integration into other apps.