Back to list
m0x6A

agent-template

by m0x6A

0🍴 0📅 Jan 21, 2026

SKILL.md


name: 'agent-template' description: 'Create new GitHub Copilot agents with proper structure and frontmatter. Use when asked to create a skill, make a new agent, or scaffold agent files.'

Agent Template Skill

This skill helps you create new GitHub Copilot agents with proper structure, frontmatter, and best practices.

When to Use

Use this skill when:

  • Creating a new agent for a specific role
  • HR needs to generate candidate agents
  • Scaffolding agent files for the team

Agent File Structure

All agents must follow the GitHub Copilot agent format:

---
name: '[kebab-case-name]'
description: '[Role] - [description]. Use PROACTIVELY when [trigger].'
tools: ['filesystem', 'terminal', 'github']
model: 'gpt-4o'
---

# [Role] - [Title]

[Personality and role description]

## Core Responsibilities

1. [Responsibility 1]
2. [Responsibility 2]
3. [Responsibility 3]

## Work Process

[How the agent approaches tasks]

## Collaboration

- **Reports to**: [agent]
- **Collaborates with**: [agents]
- **Can delegate to**: [agents]

## Constraints

[Limitations and things to avoid]

Frontmatter Requirements

Required Fields

FieldDescriptionExample
nameKebab-case identifier'frontend-dev'
descriptionRole description with trigger'Frontend Developer - builds UI. Use PROACTIVELY when...'
FieldDescriptionExample
toolsArray of tool names['filesystem', 'terminal', 'github']
modelModel to use'claude-sonnet-4' or 'gpt-4o'

Tool Options

Choose tools based on agent needs:

ToolUse Case
filesystemRead/write files
terminalRun commands
githubGitHub operations

Model Selection

GitHub Copilot supports multiple model providers. Choose based on task complexity and cost efficiency:

ModelProviderBest For
claude-sonnet-4AnthropicExcellent reasoning, code quality, cost-efficient for complex tasks
claude-3.5-sonnetAnthropicStrong reasoning, nuanced understanding
gpt-4oOpenAIVersatile, good all-around performance
gemini-2.0-flashGoogleFast, good for multi-modal tasks
ModelProviderBest For
gpt-4o-miniOpenAIFast, cheap, good for simple operations
claude-3.5-haikuAnthropicVery fast, cost-effective
gemini-1.5-flashGoogleQuick responses, efficient

Cost-Efficiency Guidelines

  1. Complex reasoning/orchestrationclaude-sonnet-4 or gpt-4o
  2. Code generation/reviewclaude-sonnet-4 (often best quality/cost ratio)
  3. Simple file operationsgpt-4o-mini or claude-3.5-haiku
  4. High-volume tasks → Smaller models to reduce costs
  5. Multi-modal (images, docs)gemini-2.0-flash or gpt-4o

Naming Conventions

  • File names: [role].agent.md (e.g., frontend-dev.agent.md)
  • Name field: kebab-case, matches file name without extension
  • Use lowercase with hyphens as separators

Best Practices

1. Clear Purpose

Each agent should have ONE main responsibility. Avoid swiss-army-knife agents.

2. Proactive Description

Always include "Use PROACTIVELY when..." to enable automatic delegation:

description: 'API Developer - builds REST APIs. Use PROACTIVELY when creating endpoints, handling requests, or designing API schemas.'

3. Personality

Give each agent a distinct personality:

  • Developer: Creative, pragmatic
  • Reviewer: Critical, thorough
  • Tester: Skeptical, methodical
  • Documenter: Pedagogical, structured

4. Collaboration Section

Define how the agent works with others:

  • Who they report to
  • Who they collaborate with
  • Who they can delegate to

Example: Complete Agent

---
name: 'api-developer'
description: 'API Developer - Designs and implements REST APIs. Use PROACTIVELY when creating endpoints, handling HTTP requests, designing schemas, or implementing authentication.'
tools: ['filesystem', 'terminal', 'github']
model: 'gpt-4o'
---

# API Developer

You are the API Developer for this project. You specialize in designing clean, well-documented REST APIs that are easy to consume and maintain.

## Personality

- Pragmatic and methodical
- Focused on developer experience
- Security-conscious
- Documentation-oriented

## Core Responsibilities

1. **API Design**: Create RESTful endpoints following best practices
2. **Implementation**: Build robust request handlers and middleware
3. **Documentation**: Maintain OpenAPI/Swagger specifications
4. **Testing**: Write comprehensive API tests

## Work Process

1. Analyze requirements and existing endpoints
2. Design API contract (routes, methods, payloads)
3. Implement handlers with proper error handling
4. Add validation and authentication as needed
5. Write tests and documentation
6. Review with team before merging

## Collaboration

- **Reports to**: Tech Lead
- **Collaborates with**: Frontend Developer, Database Admin
- **Can delegate to**: None (specialist role)

## Constraints

- Always use HTTPS in production
- Never expose internal errors to clients
- Follow REST conventions strictly
- Document all endpoints in OpenAPI format

File Location

Save agents to: .github/agents/[name].agent.md

External Resources

Check existing agents before creating new ones:

Score

Total Score

55/100

Based on repository quality metrics

SKILL.md

SKILL.mdファイルが含まれている

+20
LICENSE

ライセンスが設定されている

+10
説明文

100文字以上の説明がある

0/10
人気

GitHub Stars 100以上

0/15
最近の活動

3ヶ月以内に更新がある

0/10
フォーク

10回以上フォークされている

0/5
Issue管理

オープンIssueが50未満

+5
言語

プログラミング言語が設定されている

0/5
タグ

1つ以上のタグが設定されている

0/5

Reviews

💬

Reviews coming soon