💬 Feedback wanted: Custom agents in GitHub Copilot (partner-built & community-built) #180853
Pinned
ebndev
announced in
Copilot News and Announcements
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
-
We announced custom agents for GitHub Copilot at GitHub Universe, and now we’re launching a growing collection of partner-built agents for observability, IaC, security, databases, incident response, experimentation, automation, and more. These agents are designed to help you run real engineering workflows in your terminal, in VS Code, or on github.com — not just generate code.
This Discussion is your space to try them, break them, and tell us what’s missing. You can also learn more in our blog with a guide to get started and a full readout of our current custom agents from our partners.
🧩 What custom agents are
Custom agents are Markdown-defined domain experts that extend the Copilot coding agent. They're lightweight, zero-infrastructure, and can run wherever Copilot runs:
They help developers automate real workflows: scanning dependencies, optimizing Terraform, responding to incidents, generating release notes, tuning databases, implementing feature flags, and more.
🔌 Partner-built custom agents available today
These agents are built with GitHub partners and ready to drop into any repository (and you can find all of them here):
Observability
Security
Databases
DevOps & IaC
Incidents & Project Management
Feature Flags & Experimentation
Automation & APIs
Just drop a file in your repository:
.github/agents/<agent>.agent.mdAnd run it from Copilot CLI:
copilot --agent=<agent> --prompt "<your task>"Or select it in VS Code Copilot Chat or github.com via Copilot panel and Agents.
📣 How you can help
Please comment below with:
🙏 Thank you
Your feedback directly shapes how we extend Copilot beyond code generation into real, end-to-end engineering workflows. We can’t wait to hear what you build.
Beta Was this translation helpful? Give feedback.
All reactions