AI Pulse · Under the Radar

A tool that turns PDFs into “skills” for Claude has picked up over 1,100 stars in a short window, gaining roughly 350 stars a day, and it’s been forked by other developers already. That pace tells you people aren’t just looking at it out of curiosity. They’re building on it. And now a second project, called the-knowledge-guy, has taken the same idea further, applying it to EPUB files and letting you query an entire bookshelf at once, not just a single document.

Here’s why this matters even if you’ve never touched GitHub. A “skill” in this context is basically a packaged bit of knowledge an AI model can pull from and reason with, instead of you re-explaining context every time you open a chat. Turning a PDF or a stack of books into one of these skills means your manuals, policy documents, training material, or research library become something an AI can actually work with directly, rather than something a person has to skim and summarize first.

What’s notable is that this is no longer one clever repo that went viral and will fade. A second, more general tool showing up so quickly means a small ecosystem is forming around the same idea: feed the AI your documents in a structured way, and it becomes useful for real work instead of generic chat. For an owner, that’s a signal worth watching. If your business runs on dense documents, SOPs, contracts, technical specs, training manuals, this pattern is heading toward becoming a standard way to make that material searchable and usable by AI, not a novelty. This is the kind of thing we build into an AI command centre, turning scattered company documents into something your team can actually query and use, at /resources/omni.