Toolkit Registry Standards

Quick Reference

| Component | Purpose | Entry Trigger | |:----------|:--------|:--------------| | Skills | Knowledge injection, reusable workflows | "Create a skill for X", "Audit skill Y" | | Commands | Orchestration shortcuts, multi-skill workflows | "Create a command for X", "Audit command Y" | | Agents | Persona binding, isolated context execution | "Create an agent for X", "Audit agent Y" |

Core Standards

Component Management

For detailed specifications, see: references/full-spec.md

Key Standards:

• Skills: USE when patterns, allowed-tools, progressive disclosure
• Commands: disable-model-invocation, argument-hint, orchestration focus
• Agents: Persona binding, tool restrictions, no AskUser in workers

Validation Protocol

All components MUST pass validation before use. See references/validation.md for complete validation protocol.