| name | skill-builder |
| description | Create and manage Initra-compliant Agent Skills with proper progressive disclosure structure (Level 1-3). Use when you need to automate specialized workflows, encapsulate framework expertise, or extend agent capabilities through reusable procedural knowledge. |
🛠️ Skill Builder
This skill provides the framework for creating production-ready Agent Skills within the Initra ecosystem. It ensures all new skills follow the 3-level progressive disclosure standard for maximum token efficiency.
📋 Skill Specification
Level 1: Metadata (Always Loaded)
- name: Human-friendly display name (max 64 chars).
- description: Precise "what" and "when" clause (max 1024 chars).
- frontmatter: MUST start and end with
---.
Level 2: SKILL.md Body (Triggered)
- Overview: Brief description of the skill's goal.
- Quick Start: The 80/20 rule (simplest usage).
- Procedures: Detailed step-by-step guidance.
- Troubleshooting: Solutions for common failure modes.
Level 3: Supporting Assets (On-Demand)
- docs/: Deep reference material and ADRs.
- examples/: Concrete, runnable code patterns.
- scripts/: Executable automation for the skill.
🛠️ Execution Procedures
Step 1: Scenario Identification
Define the domain of the new skill.
Example: supabase-integration, pino-logging-standards, nextjs-proxy-management.
Step 2: Directory Scaffolding
Create the top-level structure:
mkdir -p .agents/skills/[skill-name]/docs
mkdir -p .agents/skills/[skill-name]/examples
mkdir -p .agents/skills/[skill-name]/scripts
Step 3: Drafting SKILL.md
Follow the template provided in AGENTS.md. Ensure the description uses high-density trigger words to enable autonomous matching.
🚦 Validation Checklist
📖 Reference Links