Most of us have retyped the same instructions into FlyerGPT a dozen times: the same grading rubric, the same meeting-notes summary, the same “keep this in our brand style” reminder. Three features in FlyerGPT solve exactly that kind of repetition and give ongoing work a real home base instead of scattered one-off chats: Personal Skills, Instance-Wide Skills, and Projects. Here’s what each one does and when to reach for it.
Personal Skills - Stop Retyping Your Instructions
A Skill is a saved set of instructions that FlyerGPT automatically applies whenever it’s relevant, in both ONEchat and Agent conversations. Build it once; FlyerGPT brings it back on its own after that.
- Where to build one: your name or initial → Settings → My Skills, where starter templates are available; or click the “+” button in any chat and choose Skills → Create a Skill.
- What goes into it: the instructions themselves (your formatting rules, tone, or template) plus a short “when-to-use” description that tells FlyerGPT which situations should trigger it.
- Two activation modes: Always On, applied automatically to every conversation, or Selective, applied only when the when-to-use conditions match. A visual indicator in the chat shows which Skills are currently active.
Use for: grading rubrics, turning raw advising or meeting notes into a consistent summary format, syllabus and course document formatting, or any prompt you find yourself rewriting chat after chat. Personal Skills are private to your account; up to 50 per person, and best suited to formatting, tone, and structure rather than code-execution workflows.
Instance-Wide Skills - The Same Consistency, Deployed for Everyone
Personal Skills are private, but the same idea can scale up. UDit deploys Instance-Wide Skills, and they apply automatically across every Agent and ONEchat conversation for the whole University, not just one account.
- What they’re for: standardizing brand voice, formatting, or structured output across the entire instance, so everyone gets the same consistent behavior without having to build it themselves.
- Same two modes as Personal Skills: Always On applies instance-wide with no toggling; Selective only activates under specific conditions. Either way, users see the same in-chat indicator when one is active.
- Who manages these: The AI Applications and Services Team in UDit.
If you notice FlyerGPT consistently formatting something a certain way across the University without you asking for it, that’s likely an Instance-Wide Skill at work.
Projects - A Home Base for Ongoing Work
Not every task is a one-off chat. A Project is a named workspace that keeps related chats, files, and standing instructions together for a specific initiative, so you’re not hunting through chat history or re-uploading the same documents every time you pick up a piece of work.
- Files: attach documents once at the Project level and every chat inside that Project can use them; no re-uploading per conversation.
- Instructions: up to 5,000 characters of standing guidance (tone, terminology, background context) that automatically applies to every new chat started in the Project.
- Chats: starting a new chat from the Project page keeps it in that Project’s context; it still shows up in your regular chat history, but also organized under the Project.
- Sharing: invite teammates as Editors (same access as the owner, minus deleting the Project) or Chat-Only (create and manage their own chats, view files and instructions). Each person only sees the chats they personally created inside a shared Project.
Use for: a semester-long research effort, a recurring departmental reporting cycle, or any initiative where the same background material and instructions apply across many conversations over time.
Where Projects, Agents and Skills Fit
Projects, Skills, and Agents answer three different questions. A Project answers where and why; it’s the workspace holding the goals, files, and outputs for an initiative. An Agent answers who; it’s the worker that reasons, uses tools, and takes action. A Skill answers how; it’s the reusable method that standardizes output for both. You don’t have to use all three together, but they’re built to layer: a Project can supply context to an Agent, and a Skill can standardize what either one produces.
More to Come
Please subscribe to the blog to get future posts in this series, which will cover new features as well as more tips and tricks to help you get the most out of your AI tools on campus.