it looks like we can now see reasoning going on!

Author: yeus

src/components/taskyon/TaskChainViewer.vue

@@ -51,7 +51,12 @@
     </q-tree>
     <template v-else>
       <template v-for="(task, idx) in props.selectedThread" :key="task.id">
-        <q-expansion-item v-if="reasoning.get(task.id)" label="thoughts:" dense>
+        <q-expansion-item
+          v-if="reasoning.get(task.id)"
+          label="reasoning"
+          dense
+          class="text-caption"
+        >
           <tyMarkdown :src="reasoning.get(task.id)!" />
         </q-expansion-item>
         <Task
@@ -71,6 +76,19 @@
     </template>
     <!--Render tasks which are in progress-->
     <div class="task-logs q-py-sm">
+      <tyMarkdown
+        v-if="
+          currentMessageStream?.length === 0 &&
+          currentThinkingStream &&
+          currentThinkingStream.length > 0
+        "
+        no-line-numbers
+        no-mermaid
+        :src="'THINKING:\n' + currentThinkingStream?.split('\n').slice(-20).join('\n')"
+        class="text-caption"
+        style="font-size: 0.8rem"
+      >
+      </tyMarkdown>
       <q-card
         v-if="
           !!tystate.lastTaskState.get(currentTask.id) &&
@@ -80,14 +98,10 @@
         flat
       >
         <div class="col">
-          <tyMarkdown v-if="!currentMessageStream && currentThinkingStream">
-            {{ currentThinkingStream?.slice(-200) }}
-          </tyMarkdown>
           <tyMarkdown
             v-if="currentMessageStream"
             no-line-numbers
             no-mermaid
-            :use-iframe="false"
             :src="currentMessageStream || ''"
           />
           <div>
@@ -156,11 +170,12 @@ const props = defineProps<{
 
 const reasoning = ref(new Map<string, string>())
 watch(
-  props.selectedThread,
-  (thread) => {
+  () => props.currentTask.id,
+  () => {
+    console.log('re-calculate reason lists!')
     reasoning.value.clear()
     void Promise.all(
-      thread.map(async (t) => {
+      props.selectedThread.map(async (t) => {
         const meta = await tystate.getMeta(t.id)
         if (meta) {
           const reason = getReasoning(meta)
@@ -222,9 +237,9 @@ const currentMessageStream = computed(() => {
 })
 
 const currentThinkingStream = computed(() => {
-  if (props.currentTask)
+  if (props.currentTask) {
     return streamingTracker.value.get(props.currentTask.id)?.choices?.[0]?.reasoning || ''
-  else return undefined
+  } else return undefined
 })
 
 const currentFunctionStream = computed(() => {

src/css/markdown.sass

@@ -86,7 +86,7 @@ $primary: var(--q-primary-rgb)
     border-radius: 0.25rem
     cursor: pointer
     transition: background 0.2s ease
-    z-index: 10
+    //z-index: 10
 
     &:hover
       background: rgba($secondary, 0.05)

src/modules/tools/chatCompletionTool.ts

@@ -799,12 +799,15 @@ export async function createChatCompletionTool(
       // parse the response into our own type ...
       const choice = chatCompletion?.choices[0]
 
-      let metaInfo: TaskNodeMeta = { taskPrompt: chatInfo, rawOutput: chatCompletion }
+      let metaInfo: TaskNodeMeta = {
+        taskPrompt: chatInfo,
+        rawOutput: chatCompletion,
+      }
       // get token usage for this task..
       if (currentTask && lastTaskBeforeChatCompletion) {
         if (chatCompletion) {
           console.log('save token usage...')
-          // openai & openrouter sends back the exact number of prompt tokens :)
+          // openai & openrouter  sends back the exact number of prompt tokens :)
           metaInfo = {
             ...metaInfo,
             ...(await saveTokenUsage(
@@ -820,6 +823,7 @@ export async function createChatCompletionTool(
 
           void addTaskCostInformation(chatCompletion, currentTask?.id, llmSettings, apiKeys).then(
             (newMeta) => {
+              console.log('found new task costs:', newMeta)
               void taskManager.debugDb.upsert(currentTask.id, newMeta, 'shallow_merge')
             },
           )