Standardizing the Agentic Web
SkillMD is a high-performance open-source registry built for discovering, sharing, and standardizing AI agent behaviors, custom instructions, and system prompts.
Why We Started
AI agents are moving fast, but their capabilities and prompts remain fragmented across individual repositories and workspaces. Developers spend too much time rewriting system instructions and capability schemas. SkillMD was built to make these components indexable, searchable, and reusable.
The SKILL.md Standard
We believe in the power of declarative agent specifications. By tracking and indexing SKILL.md files, we enable a world where any LLM agent can instantly read, select, and run capabilities dynamically on-demand, without recompilation.
Core Principles
Open and Independent
SkillMD is independent of Anthropic, OpenAI, or other providers. We support all models and orchestration frameworks that run on standard prompt files.
Taxonomy Driven
Instead of relying only on raw keywords, we map skills to real occupational taxonomies (like SOC Major Groups). This matches AI capabilities to real-world jobs.
Developer-Centric
Everything is designed for performance, from our instant SQLite search APIs to the clean code block copies, CLI hooks, and sitemaps.