Fabric

Fabric is an open-source framework created by Daniel Miessler in January 2024 and written in Go, designed to make AI integration practical for everyday life and work. It addresses what the project describes as AI's core challenge: not a capabilities problem, but an integration problem—making it easy to apply AI to specific, real-world tasks through a modular system of reusable prompts called Patterns. The project is MIT-licensed and hosted on GitHub, where it has accumulated over 42,000 stars as of mid-2026.

What It Is

Fabric is a command-line framework that organizes AI prompts—called Patterns—into a crowdsourced, modular library that users can apply to specific problems. Rather than opening a chat interface and writing prompts from scratch, users pipe text or URLs into Fabric and select a Pattern to process it. Patterns cover a wide range of tasks: summarizing YouTube videos, extracting key insights from podcasts, analyzing claims in articles, writing essays, explaining code, converting bad documentation into usable documentation, and generating social media posts. The framework supports dozens of AI providers natively and can be run as a REST API server, making it composable with other tools and workflows.

How the Pattern System Works

At the core of Fabric is the Pattern: a Markdown-formatted system prompt stored as a .md file. Patterns are organized by task and live in ~/.config/fabric/patterns/. Users can:

• Use built-in crowdsourced patterns from the repository
• Create private custom patterns in a separate directory that won't be overwritten by updates
• Chain patterns together via shell pipes to build multi-step AI workflows
• Set per-pattern model mappings via environment variables to route different tasks to different models

Fabric also supports prompt strategies (Chain-of-Thought, Tree-of-Thought, Chain-of-Draft, and others) that can be layered on top of any pattern using the --strategy flag.

Supported AI Providers and Deployment

Fabric supports a broad and growing list of AI backends through native integrations and OpenAI-compatible endpoints:

• Native: OpenAI, Anthropic (Claude), Google Gemini, Ollama, Azure OpenAI, Amazon Bedrock, Vertex AI, LM Studio, Perplexity, OpenAI Codex
• OpenAI-compatible: Groq, Mistral, DeepSeek, OpenRouter, Together, Venice AI, GitHub Models, DigitalOcean, Cerebras, and more

It can be installed via a one-line shell script, Homebrew (brew install fabric-ai), Winget, Scoop, or built from source with Go. Docker images are also available. A built-in REST API server (fabric --serve) exposes all core functionality over HTTP with Swagger/OpenAPI documentation, and an Ollama compatibility mode lets other tools treat Fabric as an Ollama backend.

Update: v1.4.457 (July 2026)

The latest release is v1.4.457, published July 9, 2026. Recent notable additions include:

• Claude Opus 4.7 support (v1.4.447, April 2026) with 1M-token context window
• OpenAI Codex plugin (v1.4.437, March 2026) using ChatGPT/Codex subscription OAuth
• Azure AI Gateway plugin (v1.4.417, Feb 2026) supporting AWS Bedrock, Azure OpenAI, and Google Vertex AI through a unified gateway
• Microsoft 365 Copilot integration (v1.4.380, Jan 2026) for enterprise users
• Full internationalization across 10 languages (v1.4.356, Dec 2025)
• Interactive Swagger/OpenAPI UI for the REST API (v1.4.350, Dec 2025)
• Speech-to-text support via OpenAI transcription flags (v1.4.291, Aug 2025)
• Linux ARM and Windows ARM binaries (v1.4.303, Aug 2025)

The project is actively maintained with frequent releases, reflecting strong development velocity.

Why It Matters

Fabric's design philosophy treats AI as a magnifier of human creativity rather than a replacement for it. By breaking problems into components and applying targeted prompts to each, it enables workflows that would otherwise require custom scripting or repeated manual prompting. The CLI-first design means Fabric composes naturally with Unix pipes, shell aliases, and other command-line tools—users can alias every pattern as a direct shell command, or save outputs directly to Obsidian vaults. A web interface is also available for those who prefer a GUI. The project is community-driven, with patterns contributed by users and a contributor graph showing broad participation.