Published agent returns "Your account has no workspace" since the Aug 19 release - public embed down 6 days

My published agent has been unreachable to the public since the new agent version shipped on August 19. Posting here because a support thread has been open the whole time without a substantive reply, and I cannot find anything documented about this. WHAT HAPPENS The chat widget embedded on my site loads normally - the corner bubble renders and opens - but the panel shows only: "This assistant isn't available right now." It has also lost its identity: generic "Chat" header instead of the agent name, no conversation starters, bare "Ask anything..." placeholder. Loading the underlying public agent URL directly shows the cause in plainer terms: https://us-agent.tinycommand.com/agent/<my-slug> -> "Couldn't sign in" "Your account has no workspace." The ?hide embed variant behaves identically. WHY I THINK THIS IS A ROUTING/AUTH REGRESSION, NOT MY ACCOUNT 1. An endpoint that is unauthenticated by design now returns the sign-in page instead of JSON: GET /public/agents/<slug>/info Aug 17 -> 200 application/json Aug 25 -> 200 text/html (the SPA sign-in shell) Same for /public/agents/<slug>/config. Public agent traffic is being resolved through a workspace/auth path it should not need. 2. My session authenticates fine. Loading us-agent.tinycommand.com gives me a fully working Studio - so this is not a login problem. 3. That Studio resolves to an EMPTY workspace. It redirects to /agents/new and offers to build my first agent, as if I have none. My existing agent does not appear. That looks like a workspace or asset resolution failure in the new build rather than a deleted asset. 4. A HEAD request during the app's own page load returns 503. 5. Separately: the previously working API paths (/api/knowledge, /api/agents, /api/workspaces) now return the SPA HTML shell with a 200 instead of JSON. The new build addresses endpoints via opaque path segments. Not the outage itself, but it silently breaks any external monitoring built against the documented paths - worth a changelog note for anyone else who automated against them. WHAT I HAVE RULED OUT - My embed code is unchanged and points at the correct slug. It was verified working on Aug 17 and broke on the day of the release. - I changed nothing: no embed edits, no settings changes, no republish. - Re-copying or regenerating the embed does not help - the slug is correct and the failure is downstream of it. CURRENT STATE I have switched the widget off at my site entirely, because a bubble that opens to an error reads to visitors as MY site being broken. So the site now has no chat at all. QUESTIONS FOR THE TEAM OR ANYONE ELSE AFFECTED - Is anyone else seeing published agents return "no workspace" after the Aug 19 release? - Is there a known workspace migration step customers are expected to take on the new build? - Should a PUBLIC agent route require workspace resolution at all? That seems like the root of it. Happy to supply my agent slug and account details privately to anyone from the team who picks this up.