agentskills

What problem does it solve?

Eliminates non-portable, inconsistent skill folders by standardizing the structure, metadata, and progressive loading behavior required for agent skills to be discoverable and reusable across different compliant agent hosts.

Core Features & Use Cases

• Open-standard schema and portability rules: Defines naming, directory layout, and frontmatter requirements so skills remain usable after being moved, vendored, or submoduled.
• Progressive loading architecture: Ensures agents can safely discover intent first (name/description), then load detailed instructions and resources only when needed.
• Security and supply-chain guidance: Promotes pre-flight inspection, deterministic pinning, and anti-hang/non-interactive execution patterns to reduce operational risk.
• Compliance-oriented review checklist: Provides validation criteria and references to verify that a skill meets standard formatting and safety expectations.
• Use case: When building a new capability for an AI agent, use this standard to scaffold a folder that activates reliably and behaves deterministically in non-interactive runs.

Quick Start

Use the agentskills standard to design a new skill folder by creating a SKILL.md at the root with a precise, keyword-dense description that states when the agent should and should not load it.