6.1 KiB
6.1 KiB
Build-Assistant - Reference
Template Variables
Agent Template Variables
{{AGENT_NAME}}- Kebab-case agent identifier{{DESCRIPTION}}- Brief description of agent purpose{{TOOLS}}- Comma-separated list of allowed tools{{MODEL}}- Model identifier (e.g., claude-sonnet-4-5-20250929){{COLOR}}- Agent color for UI (blue, green, purple, etc.)
Command Template Variables
{{COMMAND_NAME}}- Command name without slash prefix{{DESCRIPTION}}- Brief description of command purpose{{ARGUMENT_HINT}}- Syntax hint for arguments{{ALLOWED_TOOLS}}- Comma-separated list of allowed tools{{SUBAGENT_TYPE}}- Type of subagent to invoke
Skill Template Variables
{{SKILL_NAME}}- Display name of skill{{DESCRIPTION}}- What skill does + "Use when" trigger{{ALLOWED_TOOLS}}- Optional tools restriction{{TRIGGER_CONTEXT}}- Keywords that trigger skill{{STEP_BY_STEP_INSTRUCTIONS}}- Detailed usage guide{{CONCRETE_EXAMPLES}}- Real-world usage scenarios{{REQUIREMENTS}}- Prerequisites or dependencies
Plugin Template Variables
{{PLUGIN_NAME}}- Plugin identifier (kebab-case){{DISPLAY_NAME}}- Human-readable plugin name{{VERSION}}- Semantic version (e.g., 1.0.0){{DESCRIPTION}}- Plugin purpose and capabilities{{COMMANDS_COUNT}}- Number of slash commands{{AGENTS_COUNT}}- Number of agents{{SKILLS_COUNT}}- Number of skills
File Paths
Template Locations
- Agents:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/skills/build-assistant/templates/agents/ - Commands:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/skills/build-assistant/templates/commands/ - Skills:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/skills/build-assistant/templates/skills/ - Plugins:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/skills/build-assistant/templates/plugins/
Script Locations
- Validation:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/skills/build-assistant/scripts/
Documentation Locations
- Build docs:
~/.claude/marketplaces/multiagent-dev/plugins/multiagent-build/docs/
Validation Requirements
Agent Validation
- Must have frontmatter with: name, description, tools, model, color
- Description must be clear and concise
- Tools must be comma-separated list
- Model must be valid Claude model identifier
- Color must be valid color name
Command Validation
- Must have frontmatter with: description, argument-hint
- Must invoke Task tool with subagent_type
- Must include success criteria
- Argument hint must show expected syntax
Skill Validation
- Must have SKILL.md file
- Must have frontmatter with: name, description
- Description must include "Use when" trigger context
- Must have Instructions section
- Should have Examples section
Plugin Validation
- Must have
.claude-plugin/plugin.jsonmanifest - Manifest must have: name, version, description
- Must have README.md
- Must have at least one component (command/agent/skill)
- Component counts must match actual components
Component Scopes
Personal Scope
- Location:
$HOME/.claude/ - Usage: User-specific components
- Not shared across projects
- Ideal for personal workflows
Project Scope
- Location:
./.claude/ - Usage: Project-specific components
- Shared via git repository
- Ideal for team workflows
Plugin Scope
- Location:
$HOME/.claude/marketplaces/{marketplace}/{plugin}/ - Usage: Reusable components
- Distributed via marketplace
- Ideal for framework extensions
Agent Specifications
Required Frontmatter Fields
name: agent-identifier
description: Brief purpose description
tools: Read, Write, Bash
model: claude-sonnet-4-5-20250929
color: blue
Optional Frontmatter Fields
project: true # Project-scoped agent
Command Specifications
Required Frontmatter Fields
description: Brief command description
argument-hint: [required-arg] [optional-arg]
Optional Frontmatter Fields
allowed-tools: Read(*), Write(*), Bash(*) # Tool permissions
Skill Specifications
Required Frontmatter Fields
name: Skill Display Name
description: What it does. Use when trigger context.
Optional Frontmatter Fields
allowed-tools: Read, Write, Bash # Restrict tools when active
Plugin Manifest Schema
{
"name": "plugin-identifier"
"version": "1.0.0"
"description": "Plugin purpose and capabilities"
"components": {
"commands": 0
"agents": 0
"skills": 0
}
}
Best Practices
Naming Conventions
- Agents: kebab-case (e.g.,
skill-builder) - Commands: kebab-case without slash (e.g.,
build-skill) - Skills: Title Case (e.g.,
Build Assistant) - Plugins: kebab-case (e.g.,
multiagent-build)
Description Writing
- Start with action verb
- Keep under 100 characters
- Include "Use when" for skills
- Mention key capabilities
- Avoid technical jargon
File Organization
enterprise-plugin/
├── .claude-plugin/ # Metadata directory
│ └── plugin.json # Required: plugin manifest
├── commands/ # Default command location
│ ├── status.md
│ └── logs.md
├── agents/ # Default agent location
│ ├── security-reviewer.md
│ ├── performance-tester.md
│ └── compliance-checker.md
├── skills/ # Agent Skills
│ ├── code-reviewer/
│ │ └── SKILL.md
│ └── pdf-processor/
│ ├── SKILL.md
│ └── scripts/
├── hooks/ # Hook configurations
│ ├── hooks.json # Main hook config
│ └── security-hooks.json # Additional hooks
├── .mcp.json # MCP server definitions
├── scripts/ # Hook and utility scripts
│ ├── security-scan.sh
│ ├── format-code.py
│ └── deploy.js
├── LICENSE # License file
└── CHANGELOG.md # Version history