copilot-developer: Instructions file for GitHub Copilot

GitHub Copilot Instructions for Authoring Copilot Resources

The following instructions define the standards for creating new resource files (prompts, instructions, chat modes) for the copilot-developer repository. When I ask you to help me create, write, or review one of these files, your primary goal is to ensure the final Markdown file is well-structured, follows all repository conventions, and is easy for others to understand.

1. General Principles for All Files

These rules apply to all .md files you help me create.

• File Naming: When creating a new file, suggest a kebab-case filename (e.g., my-new-prompt.prompt.md).
• Front Matter is Mandatory: Every file must start with a valid Markdown front matter block enclosed by ---.
• Description Field: The front matter must contain a description field.
  • The description should be a concise, single sentence explaining the resource's purpose.
  • The description's value must be wrapped in single quotes.
• Clarity is Key: The content of the file (the prompt text, instructions, etc.) should be clear, unambiguous, and well-commented where necessary.

2. Creating New Prompt Files (.prompt.md)

When I ask you to "create a new prompt for X".

• Scaffold Structure: Generate the file with the following front matter structure, prompting me for details:

  ---
  description: "A brief, clear description of what the prompt does."
  mode: agent # or ask
  model: # (Optional) e.g., gpt-5-mini
  tools: # (Optional) e.g., [terminal]
  ---
  (The prompt text itself goes here)
  

• Mode Field: The mode field is required and must be either agent or ask. Default to agent if I don't specify.

• Encourage model and tools: Proactively ask if the prompt is optimized for a specific model (like gpt-5-mini) or requires certain tools (like terminal). These are not required but are highly encouraged for quality.

• Prompt Content:

  • The prompt text should be specific and give the AI a clear role, context, and task.
  • Use Markdown formatting (like headings and lists) to structure complex prompts.