Every AI tool gives you the same starting line: a blank text box. This is a business-user's guide to Fabric — Daniel Miessler's open-source library of expert prompts that someone has already refined for the job you're doing. Stop staring at the box. Start with a Pattern.
Open ChatGPT, Copilot, Gemini, Claude. What greets you is exactly what greeted you yesterday: an empty rectangle, blinking cursor, expectant silence. The tool knows how to do almost anything. It is waiting for you to know what to ask.
So you type. "Help me write…" "Can you summarize…" "I want to…" Three sentences in, you're still scaffolding the request — telling it what role to play, what tone to use, how long the output should be, what format you want, what to leave out. By the time you've described what good would look like, you've done half the thinking the AI was supposed to save you.
The problem isn't the model. The problem is that you are reinventing the prompt every single time. The same prompts, in slightly different words, written from scratch by millions of people every day. None of them written by someone who actually thought hard about how to get the best output from a model for that specific job.
Fabric is Daniel Miessler's answer to the reinvention problem. It's an open-source project on GitHub built around one simple unit: the Pattern.
A Pattern is a carefully written prompt — usually a couple of pages long, in plain markdown — that tells an AI exactly how to do one specific job. There's a Pattern for summarizing a YouTube video. A Pattern for extracting insights from an article. A Pattern for analyzing the claims in an argument. A Pattern for improving your writing. There are over two hundred of them, all visible in the repo, all readable, all free.
You don't have to install anything to use them. Open the patterns directory, click the one you want, open the system.md file, copy its contents, paste it into ChatGPT, then paste your actual input below it. That's it.
The Patterns aren't precious. They're checked into a public repo precisely so people will copy them, modify them, build their own. Daniel's whole pitch is that prompts are the fundamental unit of AI value, and the unit nobody is curating well. Fabric is the curation.
The bottleneck on good AI answers isn't the model anymore. It's whether someone has already sat down and written a really good prompt for the specific job you're trying to do.The Fabric philosophy
These are the Patterns that translate best to ordinary business work — the jobs you find yourself doing fifty times a month without thinking of them as AI jobs. Tap Copy prompt to grab the actual Pattern Daniel authored, fresh from the repo. Tap View full prompt to read what it actually says.
Patterns are fetched live from danielmiessler/fabric the moment you tap Copy or View — so you're always grabbing the latest version Daniel has published, exactly as he wrote it. All credit and licensing (MIT) sit with the original author.
The official Fabric is a command-line tool. That's how most of the YouTube demos use it, and that's why a lot of business users assume it's "for developers." It isn't. Every Pattern is just a block of plain text. If you can paste, you can use Fabric.
The flow is the same regardless of which AI tool you're in:
1. Open the Pattern in the GitHub repo (or use the Copy button above).
2. Paste it as a new message in your AI tool of choice.
3. Paste your actual input — the article, transcript, draft, whatever — right below it.
4. Hit send.
For Patterns you'll use repeatedly, save them somewhere persistent — a saved instruction, a custom GPT, a Gemini Gem, a Claude Project. Here's where in each tool:
Create a Custom GPT and paste the Pattern into the system instructions. Or use Projects — one project per Pattern works well for frequent use.
Build an agent per Pattern in Copilot Studio. Paste the system.md content into the instructions block. Pin to the sidebar for one-click access.
Each Pattern becomes its own Gem. Drop the Pattern content into the instructions field, give it the Pattern's name, save. One Gem per Pattern you use weekly.
Create one Project per Pattern, paste the Pattern into the project's custom instructions. Every chat in that project arrives with the Pattern preloaded.
If you're comfortable in a terminal, the native experience is faster. After installing Fabric (one-line installer for macOS, Linux, Windows), running a Pattern looks like:
pbpaste | fabric -p extract_wisdom
That pipes whatever's on your clipboard through the Pattern of your choice. Add aliases and you can just type extract_wisdom. Full install instructions live in the Fabric repo.
Fabric and TELOS are siblings — both Daniel Miessler's work, both trying to fix the same problem from different sides. They solve a single sentence together: the AI doesn't have what it needs to give you a useful answer.
An expertly crafted instruction for the specific job — written by someone who thought hard about how to get a model to produce a useful answer for that task.
A short structured document about you — your problems, mission, goals, strategies, projects — that any AI can load so it stops treating you like a stranger.
This is an educational page. The framework, the Patterns, the philosophy — all of it is Daniel Miessler's work. The originals are excellent, freely available, and worth your time.
The most-watched introduction to Fabric. Start here if you want to see the whole thing demonstrated end-to-end in twenty minutes.
Watch → RepositoryForty thousand stars and counting. Browse all 200+ Patterns, read the philosophy doc, fork your own.
Open on GitHub → AuthorEssays, podcasts, and the original thinking behind both Fabric and TELOS. Subscribe if you want the long arc of where this is going.
Read →