Agentic AI, MCP, and spec-driven development: Top blog posts of 2025

As the editor of the GitHub Blog, I get a front-row seat to everything that’s published here. As we wrap up 2025, I’m marking the occasion by looking back at the most popular blog posts of the year as well as some of my favorite interviews.

While AI models were the big topic of 2024, this year saw AI become your coding partner. AI agents and agentic tools were top among our most popular posts of the year.

Below, you’ll find some of our biggest announcements of the year, plus some examples of how to use these features.

Cozy up at your desk or on your couch, grab a blanket and a mug of coffee, and let’s dive in. Here’s what you read the most in 2025:

Agent mode

GitHub Copilot agent mode, announced in February, can iterate on its own code, recognize errors, and fix its mistakes in real time, right in your IDE. More simply, it’s a problem solver that understands your intent, builds a solution, and iterates until it gets it right. 

You also read all about it in our agent mode 101 guide. Here you can learn what agent mode is, how to use it, some common use cases, and how to get started.

Coding agent

In May, the GitHub Copilot coding agent was announced. Embedded directly into GitHub, the agent starts its work when you, for example, assign a GitHub issue to Copilot or prompt it in VS Code. Coding agent will explore the repository, write code, pass tests, and open a pull request for your review.

Agent HQ

Wrapping up a year of AI agents, during GitHub Universe in October, GitHub announced an ecosystem for all of your agents to work together. Agent HQ unites all of your agents on a single platform. Coding agents from Anthropic, OpenAI, Google, Cognition, xAI, and more became available directly within GitHub as part of your paid GitHub Copilot subscription.

MCP

Model Context Protocol (MCP) makes it simple for AI agents and tools to talk to each other. Each MCP server is like an ingredient in your AI stack, whether it’s Playwright for browser automation, Notion for knowledge access, or GitHub’s own MCP server with over a hundred tools. 

In April, we rolled out Agent Mode with MCP support to all VS Code users. In September, the GitHub MCP Registry launched as your new home base for discovering MCP servers, making it easier for you to build, evaluate, and even find the MCP servers you need.

Spec-driven development

Instead of coding first and writing docs later, in spec-driven development, you start with specifications, or specs. Specs become the shared source of truth. Spec Kit, our open source toolkit for spec-driven development, provides a structured process to bring spec-driven development to your coding agent workflows, making your specification the center of your engineering process.

Conversations around the fire…

Throughout the year, we brought you stories about people and projects shaping the industry. Here are some of my personal favorites and ones you can watch, read, or listen to as you curl up by the fire.

🎉 Did you know Git turned 20 this year? To celebrate, we sat down with Linus Torvalds, the creator of Git and Linux, to discuss how it forever changed software development.

🤖 It was the breach that broke the internet. Log4Shell proved that open source security isn’t guaranteed and isn’t just a code problem. Hear the untold story from Christian Grobmeier, one of the maintainers of the open source project Log4j.

🎧 Home Assistant became the most important project in so many houses. Hear about the paradox that makes Home Assistant compelling to developers from maintainer Franck Nijhof (better known as Frenck).

See you in 2026!

And that’s just scratching the surface! Explore the blog to see more exclusive interviews, career advice, open source highlights, and more of the tutorials you love from 2025.

We’ll be here next year to share more. Stay tuned—we can’t wait for you to see what’s next!

The post Agentic AI, MCP, and spec-driven development: Top blog posts of 2025 appeared first on The GitHub Blog.