Skillsbank — a skill registry for AI agents (with MCP support + public portfolios)

If you've been managing agent skills and associated scripts/references/templayes as a scattered mess of markdown files and loose scripts across machines like I have you might want to check this out. I built something to clean that up: https://aiskillsbank.com

The core idea: a central registry where you can privately or publicly store, version-control, and browse your AI skills — accessible from any machine or mobile. The MCP integration is the practical payoff: your agent can pull the latest version of any skill and all its associated files (scripts, references, templates) directly, no manual copy-paste or file sync needed.

What’s in it for you if you try it out: Public skill portfolios

You can now publicly display the skills you've built — think of it as a GitHub profile but for agent capabilities. If you're building seriously in this space, it's a way to show your work and let others browse/fork what you've created. Would love feedback on what else would make this useful for sharing skills with the community.

Half the fun has been the journey — I went deep on learning new things about agentic engineering, the associated vibe/debugging, DNS configuration, automated testing, and Claude Code in ways I never expected when I started. I've been building software with enterprise teams for 17+ years, but this is the first app I've ever started and finished/deployed all by myself (well, unless we count Claude 🤣).

Looking for honest feedback from anyone building with agents in a home lab or enterprise context:

• What's missing from your skill management workflow? • Would public portfolios be useful for your team or community? • What would actually change how you work?

Roadmap ideas I'm considering: security scanning, team sharing, skill testing/eval. Happy to discuss architecture or design decisions in the thread!