Skill Retriever semantic skill discovery for AI agents via 10K-category taxonomy (github.com)

The launch of Skill Retriever, a new plugin for Hermes Agent powered by AgentSkillOS, introduces advanced semantic skill discovery for AI agents. By filtering over 1,200 skills through a comprehensive 10,000-category taxonomy, Skill Retriever enhances the query process by providing the top five relevant skills for each user request. This capability allows AI agents to navigate a hierarchy of skills effectively, moving beyond mere keyword searches to offering functionally relevant skills, even if they appear unrelated on the surface. The integration is seamless, requiring no core modifications and incurring zero additional API costs, making it both accessible and efficient. This development is significant for the AI/ML community as it transforms skill retrieval from a flat catalog approach to a more dynamic, search-engine-like experience. By employing LLM navigation and intelligent hint injection, Skill Retriever reduces prompt bloat while maintaining a robust response mechanism, allowing for scalable performance with over 10,000 skills. Each skill is also scanned for safety, ensuring users can trust the functionality without risking security. As the demand for more sophisticated AI interactions grows, this feature positions Hermes Agent at the forefront of efficient and effective AI skill utilization.