Today, we released Skillware v0.3.6, Skillware is basiclaly an open-source framework for packaging Python capabilities (wallet screening, PDF tools, compliance checks, etc.) as skills agents call via normal tool APIs, instead of generating one-off code each time.
This release is mostly contributor-facing: CLI works out of the box (pip install skillware, python -m skillware fallback), clearer help, CI runs bundle tests under skills/, and a small wallet_screening fix, and this made me pause for a moment and be grateful of all the contributors and what we've built.
Today, we have over 40 stars and 12 contributors from all over the world. I know it might sound like nothing to some, but we started from zero, and slowly built this upwards together. I won't explain here why Skillware is better than agent skills, you can check the docs or read articles about it, but I wanna share why I am so excited about these numbers, even tho again, to some they might sound non worthy of hype.
Initially, we built skillware having machines in mind. Most people are building AI, apps, services, tools, products, targeting humans. But in 2026, AI is your core audience, the one that makes up more than half of the internet traffic. I mean, even when humans dyor they essentially employ an AI to do the search, even when you google things, AI is replying. So, the idea was to make Skillware easily accessible, discoverable, parsable, testable, and usable by LLMs and AI Agents with 0 friction.
We built the website in a way that makes it easy for agents to find, browse skills, find skills they need, and use them out of the box with one code block, no matter which model or framework they are coming from.
On top of that, our repo is not only AI friendly, but AI focused. YES! we expect that everyone is using AI tools, code assitsants and AI enabled IDEs. There is no excuse to not do it in 2026. Our repo is build in such a way, that literally any agent or model can understand it instantly and start solving issues with single shots, without back and forths, without breaking files, or ignoring important ripple effects or complementary files etc.
I take this moment to celebrate Skillware. Even if I am celebrating alone ❤️ and I would like to welcome anyone who wants to contribute to open source AI, or sharpen their python skills to join us and give it a spin, not just a project/tool, but also as modern clean cut repo tailored for agents at work. Lots of good first issues, as well as complex and sophisticated skill bundles that are not just silly md prompts.
You can check the repo at arpahls/skillware, install with pip install skillware, or visit skillware.site
Top comments (0)