
What happened
GitHub published the latest installment of its GitHub Copilot app for Beginners series, outlining four ways to tailor Copilot to a project: custom instructions, agent skills, custom agents, and MCP servers.
Why it matters
Teams that repeatedly explain conventions, run the same procedures, or connect outside tools can put that knowledge into the repository so Copilot picks it up automatically instead of being told each time.
Summaries like this, in your inbox every morning.
The guide places repository-level context at the center of working with an AI coding assistant. Its four customization types are presented as complementary: custom instructions define conventions Copilot should follow across the repository or for particular files, agent skills package repeatable procedures Copilot can load when relevant, custom agents provide focused expertise for broader tasks, and MCP servers connect Copilot to external tools, services, and data sources.
The post also points readers to Awesome Copilot, where the community has shared ready-made instructions, skills, agents, and more. For teams unsure where to start, it recommends picking one convention they keep explaining, adding it to .github/copilot-instructions.md, and seeing how Copilot applies it to the next task.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Anthropic announced dynamic workflows, a new multi-agent feature in Claude Console where the lead agent writes…

Meta released Muse Wonder in two variants

Microsoft Research found eight runs of the same AI research agent on one task plateaued at different levels, w…

Sakana AI and UC Berkeley's MASS method let one model solve tasks, judge results, and rewrite its own workflow…

Baidu's open-model Unlimited-OCR (3B total parameters, 0.5B active) reports 93.23 on the OmniDocBench v1.5 com…

Meta released Muse Wonder and Muse Wonder Full, saying the largest version beats Anthropic's Claude Fable 5.1…
