Build shared workflows for your team with Copilot Automations
By Steven Van ·
Shared automations are billed to the repository owner, act as Copilot, and are rolling out to Copilot Pro, Pro+, Max, Business and Enterprise plans.
Shared Copilot Automations are rolling out, so teams can now build and maintain AI-powered workflows together on GitHub. Until now, automations were personal: only the creator could see and edit them, and they used that person's identity and AI credits.
An automation combines prompts, triggers and tools. Copilot cloud agent runs it in the cloud, using only the tools that have been allowed. Examples from GitHub include reproducing bugs as soon as an issue is opened, capturing screenshots of changes when a pull request is marked as ready, and investigating production incidents as soon as a monitoring tool detects them.
How shared automations differ from personal ones
- Visibility and editing: everyone with read access to the repository can see a shared automation, and anyone with write access can edit it. A personal automation is visible and editable only by its creator. Runs of either type are visible to everyone with read access.
- Billing: AI credits are billed to the organization or user that owns the repository, not to the person who created the automation. GitHub Actions minutes are billed separately to the repository owner.
- Identity: actions on GitHub are attributed to Copilot rather than to one user.
- Durability: a shared automation belongs to the repository, so it keeps running if its creator leaves the team or loses access. Change history shows how its tools, triggers and prompt have evolved.
Triggers, tools and limits
Triggers include a schedule (hourly, daily or weekly), an issue being created, and a pull request being opened or synchronized. Event-based triggers can take filters, such as a search query or the files changed in a pull request. Every automation can also be started through the API, with extra context passed into the prompt.
The tools selected at creation set what Copilot can do. They can be GitHub tools or tools from MCP servers configured in the repository. GitHub tools act only in the repository the automation is scoped to, while MCP server tools can reach whatever the server's configuration and credentials allow. A Suggest tools button on GitHub can propose tools based on the prompt.
Optional AI credit limits can be set per run (minimum 30 credits), daily, weekly and monthly. Daily, weekly and monthly limits stop new runs from starting but do not interrupt a run in progress.
Safety defaults
By default, automations ignore events triggered by users without write access to the repository, to reduce prompt injection risk. This can be opted out of. If a repository requires at least one pull request approval, pull requests created by a shared automation need one additional approval by default. Administrators can remove that requirement in rulesets, but not in branch protection rules. Workflows on a pull request opened by an automation do not run until a user with write access approves them.
Availability
The rollout covers Copilot Pro, Pro+, Max, Business and Enterprise plans, and availability may vary as it progresses. Once available, shared automations can be created in any repository where the user has write access, on github.com, in the GitHub Copilot app or in Visual Studio Code. The type is chosen at creation as Shared and cannot be changed afterwards. In an organization-owned repository, an administrator must enable the "Allow use of Copilot cloud agent billed to the organization" policy.
