Fix(chat): Finalize next speaker detection logic

- Enhance `checkNextSpeaker` to handle cases where the last message is a function response or an empty model message.
- If the last message is a function response, the model should speak next.
- If the last message is an empty model message, the model should speak next.
- This ensures more robust and accurate determination of the next speaker in the conversation, completing the fix for the issue.
- Updated tests.

Fixes https://github.com/google-gemini/gemini-cli/issues/551

1 files changed, 49 insertions, 3 deletions

diff --git a/packages/server/src/utils/nextSpeakerChecker.ts b/packages/server/src/utils/nextSpeakerChecker.ts
index 3fe813db..fb00b39c 100644
--- a/packages/server/src/utils/nextSpeakerChecker.ts
+++ b/packages/server/src/utils/nextSpeakerChecker.ts
@@ -7,6 +7,7 @@
 import { Content, SchemaUnion, Type } from '@google/genai';
 import { GeminiClient } from '../core/client.js';
 import { GeminiChat } from '../core/geminiChat.js';
+import { isFunctionResponse } from './messageInspectors.js';
 
 const CHECK_PROMPT = `Analyze *only* the content and structure of your immediately preceding response (your last turn in the conversation history). Based *strictly* on that response, determine who should logically speak next: the 'user' or the 'model' (you).
 **Decision Rules (apply in order):**
@@ -65,17 +66,62 @@ export async function checkNextSpeaker(
   // that when passed back up to the endpoint will break subsequent calls. An example of this is when the model decides
   // to respond with an empty part collection if you were to send that message back to the server it will respond with
   // a 400 indicating that model part collections MUST have content.
-  const history = await chat.getHistory(/* curated */ true);
+  const curatedHistory = chat.getHistory(/* curated */ true);
 
   // Ensure there's a model response to analyze
-  if (history.length === 0 || history[history.length - 1].role !== 'model') {
+  if (curatedHistory.length === 0) {
+    // Cannot determine next speaker if history is empty.
+    return null;
+  }
+
+  const comprehensiveHistory = chat.getHistory();
+  // If comprehensiveHistory is empty, there is no last message to check.
+  // This case should ideally be caught by the curatedHistory.length check earlier,
+  // but as a safeguard:
+  if (comprehensiveHistory.length === 0) {
+    return null;
+  }
+  const lastComprehensiveMessage =
+    comprehensiveHistory[comprehensiveHistory.length - 1];
+
+  // If the last message is a user message containing only function_responses,
+  // then the model should speak next.
+  if (
+    lastComprehensiveMessage &&
+    isFunctionResponse(lastComprehensiveMessage)
+  ) {
+    return {
+      reasoning:
+        'The last message was a function response, so the model should speak next.',
+      next_speaker: 'model',
+    };
+  }
+
+  if (
+    lastComprehensiveMessage &&
+    lastComprehensiveMessage.role === 'model' &&
+    lastComprehensiveMessage.parts &&
+    lastComprehensiveMessage.parts.length === 0
+  ) {
+    lastComprehensiveMessage.parts.push({ text: '' });
+    return {
+      reasoning:
+        'The last message was a filler model message with no content (nothing for user to act on), model should speak next.',
+      next_speaker: 'model',
+    };
+  }
+
+  // Things checked out. Lets proceed to potentially making an LLM request.
+
+  const lastMessage = curatedHistory[curatedHistory.length - 1];
+  if (!lastMessage || lastMessage.role !== 'model') {
     // Cannot determine next speaker if the last turn wasn't from the model
     // or if history is empty.
     return null;
   }
 
   const contents: Content[] = [
-    ...history,
+    ...curatedHistory,
     { role: 'user', parts: [{ text: CHECK_PROMPT }] },
   ];