Hi, thanks for reaching out, yep a big issue... not only for skills but all dependencies. One of the options I see is governance... rely on a trusted listing where you with your experts curate/validate/assess and select the ones that match your quality standards. The current MCP already supports that, you just need to change the listing json file, right now i use this listing https://github.com/dmgrok/agent_skills_directory that goes get the skills to some "trusted" repos, anthropic, vercel, github.
how are you dealing with this topic at keypost.ai?
Hi HN, I built this to solve a fragmentation problem in AI agent development.
Right now, Anthropic, OpenAI, GitHub, and Vercel all publish "skills" for AI agents in separate repos with different formats. If you're building an MCP server or AI agent, you need to scrape each one individually.
This project:
- Aggregates skills from all major providers into one JSON catalog
- Auto-updates daily via GitHub Actions (versioned as vYYYY.MM.DD)
- Delivers via CDN (jsdelivr) for fast consumption
- Includes a browser at https://dmgrok.github.io/agent_skills_directory/
The catalog powers my MCP server (mcp_mother_skills) that exposes all skills to Claude, GPT, and other agents.
Technical choices:
- Pure Python (no dependencies beyond PyYAML)
- TOON format support for ~40% smaller payloads
- JSON Schema validation
- Extensible provider system (add new sources in 5 lines)
Happy to answer questions about the architecture or MCP integration patterns.
how are you dealing with this topic at keypost.ai?