fix(runtime-fallback): defer toast until dispatch succeeds, block restore of exhausted primary

packages/omo-opencode/src/hooks/runtime-fallback/auto-retry-dispatch.ts

@@ -31,10 +31,10 @@ export function createAutoRetryDispatcher(
     newModel: string,
     resolvedAgent: string | undefined,
     source: string,
-  ): Promise<void> => {
+  ): Promise<{ accepted: boolean }> => {
     if (sessionRetryInFlight.has(sessionID)) {
       log(`[${HOOK_NAME}] Retry already in flight, skipping (${source})`, { sessionID })
-      return
+      return { accepted: false }
     }
 
     const agentSettings = resolvedAgent
@@ -53,7 +53,7 @@ export function createAutoRetryDispatcher(
       if (state) {
         state.pendingFallbackPromptMayHaveBeenAccepted = false
       }
-      return
+      return { accepted: false }
     }
 
     const hadAwaitingFallbackResult = sessionAwaitingFallbackResult.has(sessionID)
@@ -62,6 +62,7 @@ export function createAutoRetryDispatcher(
     sessionRetryInFlight.add(sessionID)
     let retryDispatched = false
     let retryMayHaveBeenAccepted = false
+    let result: { accepted: boolean } = { accepted: false }
     try {
       const messagesResp = await ctx.client.session.messages({
         path: { id: sessionID },
@@ -175,14 +176,14 @@ export function createAutoRetryDispatcher(
             sessionID,
             status: reservedResult.status,
           })
-          return
+          return { accepted: false }
         }
       } else if (!isInternalPromptDispatchAccepted(promptResult)) {
         log(`[${HOOK_NAME}] Auto-retry skipped by promptAsync gate (${source})`, {
           sessionID,
           status: promptResult.status,
         })
-        return
+        return { accepted: false }
       }
       sessionAwaitingFallbackResult.add(sessionID)
       if (hadAwaitingFallbackResult) {
@@ -193,10 +194,11 @@ export function createAutoRetryDispatcher(
         state.pendingFallbackPromptMayHaveBeenAccepted = false
       }
       retryDispatched = true
+      result = { accepted: true }
     } catch (retryError) {
       if (!(retryError instanceof Error)) {
         log(`[${HOOK_NAME}] Auto-retry failed (${source})`, { sessionID, error: String(retryError) })
-        return
+        return { accepted: false }
       }
       log(`[${HOOK_NAME}] Auto-retry failed (${source})`, { sessionID, error: String(retryError) })
     } finally {
@@ -226,5 +228,6 @@ export function createAutoRetryDispatcher(
         }
       }
     }
+    return result
   }
 }

packages/omo-opencode/src/hooks/runtime-fallback/auto-retry.ts

@@ -12,8 +12,9 @@ export function createAutoRetryHelpers(deps: HookDeps) {
   const { clearSessionFallbackTimeout, scheduleSessionFallbackTimeout } = createFallbackTimeoutHelpers(
     deps,
     abortSessionRequest,
-    (sessionID, newModel, resolvedAgent, source) =>
-      autoRetryWithFallback(sessionID, newModel, resolvedAgent, source),
+    async (sessionID, newModel, resolvedAgent, source) => {
+      await autoRetryWithFallback(sessionID, newModel, resolvedAgent, source)
+    },
   )
 
   autoRetryWithFallback = createAutoRetryDispatcher(

packages/omo-opencode/src/hooks/runtime-fallback/chat-message-handler.ts

@@ -43,6 +43,7 @@ export function createChatMessageHandler(deps: HookDeps) {
     if (
       state.currentModel !== state.originalModel &&
       !state.pendingFallbackModel &&
+      !state.failedModels.has(state.originalModel) && // don't restore if original model has failed
       !isModelInCooldown(state.originalModel, state, config.cooldown_seconds)
     ) {
       const activeModel = state.originalModel

packages/omo-opencode/src/hooks/runtime-fallback/fallback-retry-dispatcher.ts

@@ -24,32 +24,45 @@ export async function dispatchFallbackRetry(
     deps.config,
   )
 
-  if (result.success && deps.config.notify_on_fallback) {
-    await deps.ctx.client.tui
-      .showToast({
-        body: {
-          title: "Model Fallback",
-          message: `Switching to ${result.newModel?.split("/").pop() || result.newModel} for next request`,
-          variant: "warning",
-          duration: 5000,
-        },
-      })
-      .catch(() => {})
-  }
-
-  if (result.success && result.newModel) {
-    await helpers.autoRetryWithFallback(
-      options.sessionID,
-      result.newModel,
-      options.resolvedAgent,
-      options.source,
-    )
+  if (!result.success || !result.newModel) {
+    log(`[${HOOK_NAME}] Fallback preparation failed`, {
+      sessionID: options.sessionID,
+      source: options.source,
+      error: result.error,
+    })
     return
   }
 
-  log(`[${HOOK_NAME}] Fallback preparation failed`, {
-    sessionID: options.sessionID,
-    source: options.source,
-    error: result.error,
-  })
+  const dispatchResult = await helpers.autoRetryWithFallback(
+    options.sessionID,
+    result.newModel,
+    options.resolvedAgent,
+    options.source,
+  )
+
+  if (deps.config.notify_on_fallback) {
+    if (dispatchResult.accepted) {
+      await deps.ctx.client.tui
+        .showToast({
+          body: {
+            title: "Model Fallback",
+            message: `Switched to ${result.newModel?.split("/").pop() || result.newModel} for next request`,
+            variant: "warning",
+            duration: 5000,
+          },
+        })
+        .catch(() => {})
+    } else {
+      await deps.ctx.client.tui
+        .showToast({
+          body: {
+            title: "Model Fallback Failed",
+            message: `Could not switch to ${result.newModel?.split("/").pop() || result.newModel}. Retry blocked.`,
+            variant: "error",
+            duration: 8000,
+          },
+        })
+        .catch(() => {})
+    }
+  }
 }