Quick Reference (30 seconds)
Core Principle: Clear separation between Agents (orchestrators) and Skills (capabilities).
Component Location Contains
Agents .claude/agents/
workflows/ (TOON + MD steps)
Skills .claude/skills/
scripts/ (UV Python) OR modules/ (single-file MDs)
Decision Tree:
Need external Python packages? → UV Script (.py) Need system commands? → UV Script (.py) Reference/documentation? → Single-File MD (.md) Simple data/rules? → Single-File MD (.md)
Naming (Mandatory Prefixes):
-
Scripts: {skill}_{action}.py (underscore)
-
Modules: {category}-{topic}.md (hyphen)
Architecture Overview
Agents = Orchestrators with Workflows
.claude/agents/{category}/{agent-name}/ ├── {agent-name}.md # Agent definition └── workflows/ # Agent-owned workflows └── {workflow-name}/ ├── WORKFLOW.toon # Structure (TOON tabular) ├── instructions.md # Overall guidance ├── checklist.md # Validation (tier-2+) └── steps/ └── step-NN-{action}.md
Skills = Capabilities with Scripts OR Modules
.claude/skills/{skill-name}/ ├── SKILL.md # Main entry (REQUIRED) ├── scripts/ # For complex automation (OPTIONAL) │ └── {skill}_{action}.py # UV scripts with prefix └── modules/ # For simple reference MDs (OPTIONAL) └── {category}-{topic}.md # Single-file MDs with prefix
When to Use This Skill
Use this skill when:
-
Creating a new skill and deciding between scripts vs modules
-
Choosing workflow complexity tier (0-3)
-
Naming new files (scripts or modules)
-
Understanding the architecture separation
-
Reverse-engineering third-party apps
Load Order: This skill should be loaded FIRST when creating any new skill, agent, or workflow.
Available Modules
Module Purpose When to Load
rule-script-vs-md.md
Decision logic for script vs MD Creating new skill content
rule-naming-convention.md
Prefix naming system Naming any new file
rule-skill-tiers.md
Tier 1-4 skill complexity Assessing skill scope
rule-workflow-complexity.md
Tier 0-3 workflow selection Creating workflows
pattern-single-file.md
Universal single-file rules Reviewing file constraints
Available Schemas
Schema Purpose When to Load
workflow-schema.toon
Canonical WORKFLOW.toon format Creating any workflow
step-schema.md
Template for step-NN-{action}.md Creating workflow steps
Decision Flowcharts
Script vs MD Decision
┌─────────────────────────────────────┐ │ NEW SKILL CONTENT NEEDED │ └──────────────┬──────────────────────┘ │ ▼ ┌─────────────────────────────────────┐ │ Requires external Python packages? │ │ (httpx, pandas, click, psutil...) │ ├─────────────┬───────────────────────┤ │ YES │ NO │ │ ↓ │ ↓ │ │ UV SCRIPT │ Requires shell/sys? │ └─────────────┴───────────┬───────────┘ │ ┌───────────┴───────────┐ │ YES │ NO │ │ ↓ │ ↓ │ │ UV SCRIPT │ MD OK │ └─────────────┴─────────┘
Workflow Tier Selection
┌─────────────────────────────────────┐ │ WORKFLOW COMPLEXITY │ └──────────────┬──────────────────────┘ │ ┌──────────┼──────────┬───────────┐ ▼ ▼ ▼ ▼ ┌───────┐ ┌───────┐ ┌─────────┐ ┌─────────┐ │Tier 0 │ │Tier 1 │ │ Tier 2 │ │ Tier 3 │ │1 step │ │3-5 │ │ 5-10 │ │ 10+ │ │hotfix │ │simple │ │ medium │ │ complex │ └───────┘ └───────┘ └─────────┘ └─────────┘
Prefix Naming Quick Reference
Scripts (UV Python)
Pattern: {skill-name}{action}.py Separator: underscore ()
Examples: kalshi_get-market.py kalshi_search-events.py system_analyze-cpu.py builder_generate-skill.py
Modules (Single-File MDs)
Pattern: {category}-{topic}.md Separator: hyphen (-)
Categories: api, pattern, rule, guide, schema, example
Examples: api-endpoints.md pattern-authentication.md rule-naming-convention.md guide-quick-start.md
Works Well With
Agents:
-
builder-skill
-
Skill creation with tier system
-
builder-workflow
-
TOON + MD workflow generation
-
builder-agent
-
Agent with workflows/ generation
Skills:
-
moai-library-toon
-
TOON format reference
-
moai-foundation-core
-
Core execution rules
-
builder-skill-uvscript
-
UV script examples
Commands:
-
/moai:1-plan
-
SPEC generation with decision framework
-
/builder:generate-skill
-
Skill creation workflow
Version: 1.0.0 Status: Active Last Updated: 2025-12-02