schemas

@MadAppGang/schemas

MadAppGang

196

19 forks

Updated 1/6/2026

View on GitHub

YAML frontmatter schemas for Claude Code agents and commands. Use when creating or validating agent/command files.

Installation

$skills install @MadAppGang/schemas

Claude Code

Cursor

Copilot

Codex

Antigravity

Details

RepositoryMadAppGang/claude-code

Pathplugins/agentdev/skills/schemas/SKILL.md

Branchmain

Scoped Name@MadAppGang/schemas

Usage

After installing, this skill will be available to your AI coding assistant.

Verify installation:

skills list

Skill Instructions

name: schemas description: YAML frontmatter schemas for Claude Code agents and commands. Use when creating or validating agent/command files.

Frontmatter Schemas

Agent Frontmatter

---
name: agent-name               # Required: lowercase-with-hyphens
description: |                 # Required: detailed with examples
  Use this agent when [scenario]. Examples:
  (1) "Task description" - launches agent for X
  (2) "Task description" - launches agent for Y
  (3) "Task description" - launches agent for Z
model: sonnet                  # Required: sonnet | opus | haiku
color: purple                  # Optional: purple | cyan | green | orange | blue | red
tools: TodoWrite, Read, Write  # Required: comma-separated, space after comma
skills: skill1, skill2         # Optional: referenced skills
---

Field Reference

Field	Required	Values	Description
`name`	Yes	`lowercase-with-hyphens`	Agent identifier
`description`	Yes	Multi-line string	3-5 usage examples
`model`	Yes	`sonnet`, `opus`, `haiku`	AI model to use
`color`	No	See colors below	Terminal color
`tools`	Yes	Tool list	Available tools
`skills`	No	Skill list	Referenced skills

Color Guidelines

Color	Agent Type	Examples
`purple`	Planning	architect, api-architect
`green`	Implementation	developer, ui-developer
`cyan`	Review	reviewer, designer
`orange`	Testing	test-architect, tester
`blue`	Utility	cleaner, api-analyst
`red`	Critical/Security	(rarely used)

Tool Patterns by Agent Type

Orchestrators (Commands):

Must have: Task, TodoWrite, Read, Bash
Often: AskUserQuestion, Glob, Grep
Never: Write, Edit

Planners:

Must have: TodoWrite, Read, Write (for docs)
Often: Glob, Grep, Bash

Implementers:

Must have: TodoWrite, Read, Write, Edit
Often: Bash, Glob, Grep

Reviewers:

Must have: TodoWrite, Read
Often: Glob, Grep, Bash
Never: Write, Edit

Command Frontmatter

---
description: |                 # Required: workflow description
  Full description of what this command does.
  Workflow: PHASE 1 → PHASE 2 → PHASE 3
allowed-tools: Task, Bash      # Required: comma-separated
skills: skill1, skill2         # Optional: referenced skills
---

Field Reference

Field	Required	Values	Description
`description`	Yes	Multi-line	Command purpose and workflow
`allowed-tools`	Yes	Tool list	Tools command can use
`skills`	No	Skill list	Referenced skills

Validation Checklist

Agent Frontmatter

Command Frontmatter

Common Errors

Invalid YAML Syntax

# WRONG - missing colon
name agent-name

# CORRECT
name: agent-name

Incorrect Tool Format

# WRONG - no spaces after commas
tools: TodoWrite,Read,Write

# CORRECT
tools: TodoWrite, Read, Write

Missing Examples

# WRONG - too generic
description: Use this agent for development tasks.

# CORRECT
description: |
  Use this agent when implementing TypeScript features. Examples:
  (1) "Create a user service" - implements service with full CRUD
  (2) "Add validation" - adds Zod schemas to endpoints
  (3) "Fix type errors" - resolves TypeScript compilation issues

More by MadAppGang

View all

external-model-selection

1,265

Choose optimal external AI models for code analysis, bug investigation, and architectural decisions. Use when consulting multiple LLMs via claudish, comparing model perspectives, or investigating complex Go/LSP/transpiler issues. Provides empirically validated model rankings (91/100 for MiniMax M2, 83/100 for Grok Code Fast) and proven consultation strategies based on real-world testing.

claudish-usage

247

CRITICAL - Guide for using Claudish CLI ONLY through sub-agents to run Claude Code with OpenRouter models (Grok, GPT-5, Gemini, MiniMax). NEVER run Claudish directly in main context unless user explicitly requests it. Use when user mentions external AI models, Claudish, OpenRouter, or alternative models. Includes mandatory sub-agent delegation patterns, agent selection guide, file-based instructions, and strict rules to prevent context window pollution.

model-tracking-protocol

196

MANDATORY tracking protocol for multi-model validation. Creates structured tracking tables BEFORE launching models, tracks progress during execution, and ensures complete results presentation. Use when running 2+ external AI models in parallel. Trigger keywords - "multi-model", "parallel review", "external models", "consensus", "model tracking".

xml-standards

196

XML tag structure patterns for Claude Code agents and commands. Use when designing or implementing agents to ensure proper XML structure following Anthropic best practices.