Users report Claude Tag keeps defaulting to non-threaded responses even after being told to use threads, claiming it already has that in memory. A team member confirmed they're looking into it and reached out directly for follow-ups.
A team member is investigating a pattern where Claude Tag repeatedly ignores instructions to respond with threaded comments, noting the issue appears more prevalent with Sonnet 5 than other models.
Claude Tag's overly eager behavior in Slack is being addressed with a fix in progress. In the meantime, users can disable 'Respond Automatically' via the Configure link so it only replies when explicitly tagged.
Claude Tag is responding to all Slack messages as if they're direct instructions. A fix is in progress; the interim workaround is toggling off 'Respond automatically' so it only activates when explicitly tagged.
Claude managed agents offer the same core benefits as the Agent SDK but with sandboxing and infrastructure handled on Anthropic's side, making it a lower-overhead option for web-based agent builders.
