THE SHORT ANSWER
The update centralizes debugging in Team Test by displaying all client logs in one window and introducing LogService:GetLogHistoryAsync for script access, limited to 512 messages per user.
- Team Test logs now appear in the local Output window with a sources dropdown.
- New LogService:GetLogHistoryAsync API allows scripts to read collaborator and server logs.
- Server logs require the User.fromId(0) parameter.
- Only 512 recent messages are available per user.
- The feature is restricted to Team Test sessions and does not apply to live games.
Roblox Studio is adding a feature to its Team Test mode that lets developers see error logs from collaborators without leaving the window. Previously, debugging required asking a teammate to copy-paste errors or take screenshots. The new update centralizes this information in the standard Output window.
The primary change affects the User Interface. When a developer starts a Team Test session, the Output window gains a new context filter. This filter includes a "Sources" section that lists every participant in the session. Selecting a specific name filters the output to show only that user's logs. The system prefixes these messages with the user's name, such as [builderman] Loaded 12 enemies.
Developers can also access this data programmatically. The update introduces a new method, LogService:GetLogHistoryAsync. This function allows scripts to read the log history of a collaborator or the Team Test server. To read the server's logs, a script must pass User.fromId(0) as the user parameter. The method returns a list of tables containing the message, message type, and timestamp.
The implementation has specific limitations. The API only works during Team Test sessions. It is disabled for local solo play and live games. Additionally, the method returns only the most recent messages, with a hard cap of 512 entries per user. This matches the limit of the existing LogService:GetLogHistory method.
Scripters must handle errors, as the network call can fail. The official documentation recommends wrapping the call in a pcall block. The feature also supports AI agents through the Studio MCP, allowing them to read logs while debugging a live session.