feat: implement notification thread management #12
Reference in New Issue
Block a user
Delete Branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Context
Currently the MCP server can list notifications and mark all as read, but cannot
interact with individual notification threads — marking a single one as read,
fetching its details, or checking whether there are any unread notifications at all.
Endpoints to implement
/notifications/newcheck_notifications_new/repos/{owner}/{repo}/notificationslist_repo_notifications/repos/{owner}/{repo}/notificationsmark_repo_notifications_read/notifications/threads/{id}get_notification_thread/notifications/threads/{id}mark_notification_thread_readAcceptance Criteria
check_notifications_newreturns{"has_new": true/false}(wraps the API's integer response)list_repo_notifications(owner, repo, all, limit, page)mirrorslist_notificationsinterfacemark_repo_notifications_read(owner, repo)returns{"success": true}get_notification_thread(id)returns full thread objectmark_notification_thread_read(id)returns{"success": true}on 205Technical Notes
GET /notifications/newreturns{"new": 0}— a count, not a boolean; wrap it:{"has_new": result["new"] > 0, "count": result["new"]}PATCH /notifications/threads/{id}returns 205 Reset Content (not 204) — handle in_reqor inlinelist_repo_notificationssupportssinceandbeforedatetime params (ISO 8601) — add them as optional args