Show HN: A GitOps-style registry for AI agent Workflows, Skills and MCP servers (github.com)

A new GitOps-style registry for AI agents has been introduced, aimed at optimizing dynamic capability routing for AI workflows by loading only the necessary skills, work processes, and multi-connector platform (MCP) instructions as required by individual tasks. This experimental repository enhances how capabilities are managed by treating them as versioned infrastructure rather than a static collection, allowing agents to efficiently resolve user requests based on context-specific criteria like task and role. Significantly, this registry promotes modular, reproducible capability management for teams operating multiple agents across various tasks. By reducing the bloated capability setups that often strain resources, it allows for a more streamlined approach where agents can either dynamically access necessary components or derive minimalist setups for specific roles or tasks. Key functionalities include explicit Git tracking of enabled capabilities, a clear separation of trusted and experimental capabilities, and safer runtime environments for MCP servers. Overall, this initiative represents a crucial step toward making AI agent configurations more efficient and manageable, addressing common challenges in AI capability sprawl.