ui: Add a Desktop surface for the sandbox's Computer Use desktop (VNC)

Cargo.toml

@@ -77,6 +77,8 @@ tokio-util = "0.7"
 libc = "0.2"
 gethostname = "1"
 tokio-tungstenite = { version = "0.24", features = ["rustls-tls-webpki-roots"] }
+# The Desktop surface: RFB client over the sandbox's noVNC websocket.
+vnc-rs = "0.5"
 futures = "0.3"
 # JPEG decode for the browser screencast; the same crate gpui already builds.
 image = { version = "0.25", default-features = false, features = ["jpeg"] }

crates/keiki-api/src/lib.rs

@@ -78,6 +78,7 @@ pub enum ConversationAction {
     Messages,
     Steer,
     Terminal,
+    Desktop,
     Browser,
     /// Ends an inter-agent (`agent:`) thread — not a contact block.
     End,
@@ -92,6 +93,7 @@ impl ConversationAction {
             Self::Messages => "messages",
             Self::Steer => "steer",
             Self::Terminal => "terminal",
+            Self::Desktop => "desktop",
             Self::Browser => "browser",
             Self::End => "end",
         }
@@ -105,6 +107,15 @@ pub struct TerminalSize {
     pub rows: u16,
 }
 
+/// A way into the sandbox's desktop: its noVNC websocket behind the
+/// provider's preview proxy, plus the token that proxy wants
+/// (`x-daytona-preview-token`). Short-lived; fetch one per connection.
+#[derive(Debug, Clone, Deserialize)]
+pub struct DesktopViewer {
+    pub url: String,
+    pub token: String,
+}
+
 /// A short-lived capability to watch and drive the browser the conversation's
 /// agent handed off: `url` is the runner's screencast WebSocket, `token` the
 /// first message's credential. The bearer secrets stay on the platform.
@@ -1068,6 +1079,29 @@ impl Client {
         Ok(response.available)
     }
 
+    /// A viewer for the conversation sandbox's desktop (Computer Use).
+    /// Offered exactly where [`Self::terminal_available`] is (same sandbox),
+    /// but that answer comes from the platform's records — `None` here is
+    /// the provider's word that the sandbox is gone.
+    pub async fn open_desktop(
+        &self,
+        access_token: &str,
+        locator: &ConversationLocator,
+    ) -> Result<Option<DesktopViewer>, Error> {
+        let request = self
+            .http
+            .post(self.conversation_endpoint(locator, Some(ConversationAction::Desktop))?)
+            .bearer_auth(access_token);
+        match self.send_json(request).await {
+            Ok(viewer) => Ok(Some(viewer)),
+            Err(Error::Api {
+                status: StatusCode::NOT_FOUND,
+                ..
+            }) => Ok(None),
+            Err(error) => Err(error),
+        }
+    }
+
     /// Open a shell in the conversation's sandbox; returns the terminal id.
     pub async fn open_terminal(
         &self,

crates/ui/Cargo.toml

@@ -22,9 +22,11 @@ gpui.workspace = true
 gpui_platform.workspace = true
 gpui_tokio.workspace = true
 tokio.workspace = true
-futures.workspace = true
 tokio-tungstenite.workspace = true
+url.workspace = true
+vnc-rs.workspace = true
 image.workspace = true
+futures.workspace = true
 reqwest.workspace = true
 async-trait.workspace = true
 serde.workspace = true