SkillCenter: Large Source-Grounded Skill Library for AI Agents

Autonomous AI agents are designed to execute complex tasks with minimal human oversight, but they often struggle with a lack of robust, grounded operational knowledge. This deficiency can lead to outputs that are not only incorrect but also insecure or difficult to maintain. To address this, a new resource called SkillCenter has been developed. SkillCenter is presented as the largest open-source skill library for AI agents, boasting over 216,000 structured skills across 24 domain bundles. A key feature is its "source-grounded" skills, which account for more than half of the library. These skills are meticulously extracted from peer-reviewed journals, arXiv, and over 24,000 technical sources, ensuring that each claim can be traced back to an exact quotation in its original source. The library's creation involves an end-to-end framework that includes multi-source acquisition, an LLM-based quality gate (SkillGate), template-driven generation, iterative source-grounding, and quality-controlled publishing. This rigorous process aims to provide agents with not just executable, but also correct, secure, and maintainable operational knowledge, all packaged as offline-searchable SQLite FTS5 bundles.