You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Host support for persistent (pip/pinned) app views: clinical use case where inline rendering breaks the core loop #754
We build an OCD treatment program (Conicia) and ship an MCP app: swipeable lesson cards patients work through inside the chat, with a conversational layer around them (the assistant discusses the material, and a guardrail routes compulsive reassurance-seeking to in-the-moment tools).
Inline rendering breaks the core loop. The patient reads a card, asks the assistant a question, and after two exchanges the deck has scrolled off-screen. Re-invoking the tool stacks duplicate decks. Clinically, the conversation around the material is the entire reason to host lessons in chat, so "either chat or view the app, not both" defeats the point.
What we have tried:
Stable view identity: returning a constant viewUUID in _meta on repeat tool calls, hoping hosts route the result to the existing view rather than instantiating a new one. Host behavior appears to be per-call instantiation regardless.
`requestDisplayMode("fullscreen")": works where supported, but fullscreen suspends the conversation, which is the opposite of the goal.
The ask: host support for the persistent/pinned direction the spec already gestures at (pip), or letting app views open in the persistent side panel some hosts already have for their own artifacts. A view that stays put while the transcript scrolls would fix this class of problem completely, for us and for any app whose view is a workspace rather than a one-shot result (maps, dashboards, documents).
Happy to be a design partner or demo the clinical use case.
I built Turnfeed, a multi-user social feed inside ChatGPT, and this maps closely to the feed-as-workspace problem. Would a host-owned logical app instance that can reattach a fresh renderer be enough for Conicia, or do you need the exact view to remain alive while the conversation continues?
Good question, and the answer splits by surface for us.
For the lesson deck, a host-owned logical instance that reattaches a fresh renderer would be enough. The deck state is just a card index, so we can serialize and restore it and a remount would be invisible. What we need there is the routing: the same logical instance on repeat tool calls instead of a new deck stacked below the last one.
For our in-the-moment practices it is not enough. Those are timed, animated runs the patient is doing right then, some with audio (urge surfing, a defusion practice). A remount restarts the practice at exactly the wrong moment, and audio would cut. There we need the exact view to stay alive and keep running while the conversation continues around it.
So: a reattachable logical instance fixes the duplication problem, and a persistent live view fixes the workspace problem. We need the second one, and we would take the first as a real improvement in the meantime.
Activity
TheodorNEngoy commentedon Aug 29, 2026
I built Turnfeed, a multi-user social feed inside ChatGPT, and this maps closely to the feed-as-workspace problem. Would a host-owned logical app instance that can reattach a fresh renderer be enough for Conicia, or do you need the exact view to remain alive while the conversation continues?
Conicia commentedon Sep 2, 2026
Good question, and the answer splits by surface for us.
For the lesson deck, a host-owned logical instance that reattaches a fresh renderer would be enough. The deck state is just a card index, so we can serialize and restore it and a remount would be invisible. What we need there is the routing: the same logical instance on repeat tool calls instead of a new deck stacked below the last one.
For our in-the-moment practices it is not enough. Those are timed, animated runs the patient is doing right then, some with audio (urge surfing, a defusion practice). A remount restarts the practice at exactly the wrong moment, and audio would cut. There we need the exact view to stay alive and keep running while the conversation continues around it.
So: a reattachable logical instance fixes the duplication problem, and a persistent live view fixes the workspace problem. We need the second one, and we would take the first as a real improvement in the meantime.