TL;DR >> 113 patterns collected from real production systems. From Plan-Then-Execute to Swarm Migration, learn what actually works when building AI agents that ship. <<
Something happened over the 2025 winter holidays that caught everyone by surprise. While people were supposed to be relaxing with family and exchanging gifts, a quiet revolution was underway—and it showed up in the metrics.
# The Christmas That Changed Everything
The GitHub repository for “Awesome Agentic Patterns” had been growing steadily since its launch. But around Christmas, the growth chart went vertical. In just a few days, the repository jumped from relative obscurity to nearly 2,500 stars. The website traffic mirrored this spike. Something had clicked.
But the real story wasn’t in the metrics—it was in who was talking about AI agents.
When the Legends Went Public
Linus Torvalds, creator of Linux and Git, wrote about using AI coding agents for “vibe coding” and programming guitar pedal effects. Think about that for a second. The person who literally invented the version control system that powers modern software development was publicly embracing agents.
Tobias Lütke, CEO of Shopify and already deep into agent-assisted development, declared it his “most productive time.” This from someone running one of the world’s largest e-commerce platforms.
Perhaps most telling was Armin Ronacher, creator of Flask—one of the most respected voices in Python. He had been skeptical of coding agents, publicly raising concerns about their limitations. Then, seemingly overnight, his stance shifted. He started promoting agent-assisted workflows, documenting his learnings, and acknowledging that the technology had crossed a threshold.
The Real Bottleneck: Time
... continue reading