agent-md

Standardize agent persona file syntax and architecture across GitHub Copilot, OpenCode, and Claude Code.

3|1|Updated Apr 12, 2026
One-click install
npx skills add https://github.com/Cogni-AI-OU/cogni-ai-agent-skills --skill agent-md
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-md
Source: https://github.com/Cogni-AI-OU/cogni-ai-agent-skills/tree/main/agent-md
Command: npx skills add https://github.com/Cogni-AI-OU/cogni-ai-agent-skills --skill agent-md

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

agent-md prevents misconfigured agent persona files by clarifying the required YAML frontmatter and platform-specific Markdown structure across GitHub Copilot, OpenCode, and Claude Code.

Core Features & Use Cases

  • Schema & Structure Reference: Establishes the correct frontmatter fields and body layout for custom agent persona files like *.agent.md, CLAUDE.md, and opencode.json.
  • Agents vs Skills vs Instructions Boundary: Guides authors to separate persona (“Who”) from execution playbooks (“How”) and domain rules (“Rules”) to reduce duplication and context drift.
  • Guardrails for Safety & Portability: Covers negative-scope limitations, boundary patterns (Always/Ask first/Never), size-limit considerations, and relative-link portability.

Quick Start

Use the agent-md skill to validate your custom persona file named my-agent.agent.md before deploying it to Copilot, OpenCode, or Claude Code.

Frequently Asked Questions about agent-md

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I format YAML frontmatter for agent persona files?▼

YAML frontmatter for agent persona files requires specific schema fields and a structured Markdown body to prevent scope confusion across GitHub Copilot, OpenCode, and Claude Code. Establishing correct frontmatter ensures platforms parse permissions and persona definitions accurately.

What is the difference between Agents, Skills, and Instructions in agent persona files?▼

Agents, Skills, and Instructions separate persona responsibilities in agent files: Agents define "Who" the AI is, Skills define "How" it executes tasks, and Instructions define the domain "Rules" it follows. Separating these boundaries reduces context drift and duplication.

Does GitHub Copilot support custom agent persona files with tool permissions?▼

GitHub Copilot supports custom agent persona files like *.agent.md, but requires respecting platform-specific constraints for tool permissions and size limits. Validating schema correctness and boundary patterns ensures portability across Copilot, OpenCode, and Claude Code.

How do I validate CLAUDE.md and opencode.json for platform compatibility?▼

Validating CLAUDE.md and opencode.json for platform compatibility involves checking schema correctness, YAML frontmatter fields, boundary patterns, and relative-link portability. Correct validation prevents misconfigured tool permissions and size-limit violations across different coding platforms.

What are the limitations of using relative links in agent persona Markdown files?▼

Relative links in agent persona Markdown files face portability limitations across GitHub Copilot, OpenCode, and Claude Code due to differing platform constraints and size limits. Enforcing relative-link portability and negative-scope boundaries prevents broken references and scope confusion.