Linear
You choose per team what agents may do. What is allowed on a team also covers its issues and its sub-teams. Minerva reaches the teams the connected Linear account can see, and asks Linear for read access at first and for access to write once you allow an action that writes.
Permissions
Section titled “Permissions”| Permission | What agents can do |
|---|---|
| Read issues | List teams, read a team’s statuses, labels and members, list and search its issues, and read an issue with its sub-issues and comments. |
| Comment on issues | Comment and reply to comments. |
| Create issues | Create issues. |
| Edit issues | Change an issue’s title, description, status, priority, labels, assignee and due date. |
Risks to weigh
Section titled “Risks to weigh”- Agents act as you. Comments, new issues and changes appear under your name, and the people involved are notified as they would be for your own changes.
- Linear’s automations run. Triage rules and integrations may act on what agents change. Minerva does not control them.
- A team covers its sub-teams, including sub-teams added later.
- Comments and new issues are a way out. Everyone who can see the team reads them, and they can contain anything the agent read.
Minerva refuses a status change when Linear could then close or reopen issues in other teams on its own. Where Linear does not show a team’s whole chain of parent teams, a block on an action anywhere in the connection also blocks it there.
What agents cannot do
Section titled “What agents cannot do”Agents cannot use projects, cycles, documents or attachments, move issues between teams, or delete or archive issues. Links to Linear in what they write may point only to issues in the same team. They do not see the titles of other issues, projects and documents in links.