Search the guide

Type to search page titles, headings, and content.

Use ↑ ↓ to select, Enter to open, and Esc to close.

Enterprise governance playbook

Locking Down
GitHub Copilot

A practical guide to the controls, policies, and audit layers that keep Copilot useful, secure, and aligned with your organization.

Start with the mental model Kept current automatically

Where controls live

Layer Examples Applies to
Enterprise/Org Policies (“AI controls”) Feature enable/disable, model access, cloud agent per-org allow Whatever surfaces the policy covers
managed-settings.json MCP allow/deny, marketplaces, plugins, sandbox, permissions CLI, VS Code, Copilot app, cloud agent, JetBrains
Content controls Content exclusion, public-code matching Suggestion/context generation
Network controls IP allow lists, subscription routing, cloud agent firewall Access to GitHub/Copilot infra
Visibility/audit Audit log, streaming, usage metrics Governance/monitoring, not prevention

Golden rule for conflicts: enterprise beats org; across multiple orgs in one enterprise, least-restrictive usually wins except for a sensitive shortlist; across multiple enterprises, most-restrictive wins.

Sections

How this site stays current

A scheduled agentic workflow runs daily (hourly during confirmed GitHub/Microsoft live events), reviews the GitHub Blog changelog and GitHub Copilot documentation, and opens a pull request when governance guidance needs updating or a new control has shipped.