What if AI already knew how you work?
Introducing Skills in Friday AI
AI tools are good at a lot of things. Remembering how you work is not one of them.
Each conversation starts from zero. If you care about the tone of the content, you restate it. If you have standards, you explain them again. If your team works a certain way, you re-type it.
One Friday user put it plainly: “I’d love if I could just get Friday to follow me from meeting to meeting and learn about how I do things.”
The model can follow instructions. It just does not carry them forward.
Skills are how we solve that.
What a skill actually is
A skill is a reusable set of instructions that you can apply to any work AI does for you. Under the hood, it’s a SKILL.md file: a markdown document containing workflows, behavioral rules, preferences, or best practices that define how AI should work.
The beauty of skills is that they don’t exist in a closed ecosystem. You can use skills developed by tool creators themselves, like the official Svelte skill from the Svelte team, explore curated libraries like skills.sh, or write your own.
In Friday, that means you don’t have to work from scratch. You can grab your favorite community developed skill, pull it into your private instance, and immediately see Friday apply them to the work you want done.
What skills do for you
Before getting into examples, it helps to name what’s actually happening when a skill is installed:
Skills are automatically applied to your work. Once added, Friday picks them up across conversations and workspaces without you restating them. When work starts, relevant skills are applied by default, or you can choose specific ones during setup.
You get consistent execution every time. Skills act as a standard. Friday doesn’t reinterpret tasks from scratch. It follows the same instructions, so outputs stay predictable.
You can use existing skills or create your own. Install pre-built skills or define one in conversation. No setup or technical work required.
They evolve as you work. You can edit and refine skills anytime. Over time, they better match how you want things done.
How skills works in Friday
To add skills to Friday, you can add a skill in Friday using two ways:
If a skill already exists, paste its URL into a Friday conversation with a line of context about what you’re using it for. Friday will read it, confirm what it does, and apply it going forward. Skills from skills.sh work this way, as do skills shared directly by framework maintainers or other teams.
If you want one built around how you specifically work, start a conversation and ask Friday to create it. Friday will ask a few questions to understand what the skill should do, how it should behave, and when it should apply. It can ingest source material — documentation, writing samples, code — and turn that into a reusable capability that runs consistently from that point on.
Once a skill is added, Friday cross-references it against your other conversations and workspaces without you having to wire anything up.
How our team uses skills internally
Everyone in our company uses skills in one way or another. They’ve turned into a huge contributing factor of working effectively and efficiently.
For example, every piece of content we write (including this article) runs through a copywriting skill that edits checking for the tone, structure, and voice of Friday AI, our brand. Rather than imbuing each ask with the same set of context again and again and again, those requirements now live in a skill. It carries those preferences when reviewing anything generated on company pages.
That skill looks something like this:
This skill was built inside Friday itself. We shared previous pieces of our team’s writing alongside some aspirational examples of what we hoped to sound like. We added a few hard rules, including no use of the dreaded em dash, and answered additional follow-up questions. Friday generated the skill from that material.
Now when our team writes new content, we can focus on the quality of the idea and story rather than re-litigate tone or voice.
For more examples of skills our team uses regularly, including favorites for brainstorming, debugging, and writing code in unfamiliar frameworks, we wrote about them here: https://medium.com/friday-ai/841623598b7e
What skills are actually doing
The through-line across all of these is that they encode a way of working. Preferences that would otherwise have to be re-stated. Standards that would otherwise have to be re-explained. Processes that would otherwise exist only in your head.
The more skills you add to Friday, the more it reflects how you actually think. Not approximately. Exactly. Your voice in writing. Your standards for code. Your process for working through problems.
The interesting territory ahead is what happens when a team’s collective knowledge starts to live in skills rather than in documents no one reads or onboarding sessions no one remembers. We’re still figuring that part out.
For now, the simplest version: you stop explaining yourself from scratch every time, and the AI stops acting like it has no idea who you are.
Try it now on hellofriday.ai.
This article was originally published on March 26, 2026 on Medium.