How We Got LLMs to Recommend Our Open Source Library — Christopher Burns, Inth
Christopher Burns from Inth describes how his open source library c15t (a consent banner) began being recommended by AI agents — from April 13th, this became the largest source of new users.
To make his project AI-friendly, he built a documentation pipeline that generates files specifically for agents. His main insights: write llms.txt by hand instead of generating it, serve markdown instead of HTML, and bundle documentation directly in the package because agents typically read source code and node_modules rather than visiting websites. He saved nearly half the tokens agents need by sending bundled markdown inside the package.
Open on YouTube →
Vibekollen prepared this summary with AI from the original publication. The content belongs to AI Engineer.
