Organization Level Instructions
GitHub Copilot now supports organization-level instructions, allowing administrators to define guidance that applies to all repositories and all users within a GitHub organization. This is a powerful tool to promote consistency, security, and governance across all AI-assisted development workflows.
This page explains how org-level instructions differ from project-level .copilot-instructions.md files and how to use each effectively.
What Are Organization-Level Instructions?
Organization-level instructions are set in GitHub settings (via the Copilot tab) and are automatically injected into every Copilot Chat interaction for members of the organization — regardless of which repository they are working in.
This creates a baseline AI behavior model for all teams under the org umbrella.
How It Differs from Project-Level Instructions
| Feature | Organization-Level | Project-Level |
|---|---|---|
| Scope | All users & repos in the org | Only applies to a specific repo |
| Who manages it | Org admins | Repo maintainers or developers |
| When it loads | Always in context | Only when working in that repo |
| Purpose | Company-wide guidance, policies, tone | Repo-specific architecture, stack, naming, etc. |
| Format | Text box in GitHub settings | Markdown file in repo root: .github/copilot-instructions.md |
These two types of instructions are complementary: org-level defines global rules, project-level adds local specificity.
Use Cases for Organization-Level Instructions
- Security and Privacy Rules
Never suggest using secrets or API keys directly in code.
Avoid using eval() or direct SQL string construction.
- Documentation and Learning Resources
When asked about frontend theming, refer to the Confluence Docs at <name>.
Link to internal API documentation for auth-related questions.
- Consistency Across Teams
Use PascalCase for class names and camelCase for functions.
Always wrap DB calls with the internal SafeQuery abstraction.
- Style and Formatting Rules
All logs must use LoggerService.debug() — never console.log().
Use ?? over || for nullish checks.
- Process Guidelines
For any deployment questions, remind the user to check the InfraRunbook first.
When unsure about security decisions, suggest reaching out in #ask-security.
Best Practices for Writing Org-Level Instructions
- Be clear and prescriptive: Write them like onboarding rules — not suggestions.
- Avoid repo-specific logic: Don’t reference repo file paths or local variables.
- Use structured categories: Break into sections like “Security”, “Logging”, “Naming”, etc.
- Keep it short: Aim for < 1,000 words to avoid hitting context compression limits.
- Update periodically: Sync changes with team-wide rollouts, language style updates, or new security policies.
When to Use Project vs Org Instructions
| Scenario | Use This |
|---|---|
| Defining architecture rules for a monorepo | Project-level |
| Standardizing logging across all projects | Organization-level |
| Enforcing naming conventions per team | Project-level |
| Controlling AI suggestions for secret usage | Organization-level |
| Teaching a repo-specific design pattern | Project-level |
How to Configure
- Go to your GitHub organization settings.
- Navigate to the Copilot tab.
- Click Custom Instructions.
- Enter your organization-wide instructions in the editor box.
- Save — they’re now live for all Copilot Chat interactions in the org.
Organization instructions are your first layer of LLM governance. They establish shared values and behaviors that AI should reflect — turning Copilot into an extension of your engineering culture.