Replace thinking with native Gemini thinking summaries (#400)

This uses Gemini's native [thinking
summaries](https://cloud.google.com/vertex-ai/generative-ai/docs/thinking#thought-summaries)
which were recently added to the API.

Why? The grafted thinking would sometimes cause weird issues where the
model, especially Gemini 2.5 Flash, got confused and put dyad tags like
`<dyad-write>` inside the `<think>` tags.

This also improves the UX because you can see the native thoughts rather
than having the Gemini response load for a while without any feedback.

I tried adding Anthropic extended thinking, however it requires temp to
be set at 1, which isn't ideal for Dyad's use case where we need precise
syntax following.

src/ipc/utils/get_model_client.ts

@@ -116,6 +116,7 @@ export async function getModelClient(
         ? createDyadEngine({
             apiKey: dyadApiKey,
             baseURL: dyadEngineUrl ?? "https://engine.dyad.sh/v1",
+            originalProviderId: model.provider,
             dyadOptions: {
               enableLazyEdits: settings.enableProLazyEditsMode,
               enableSmartFilesContext: settings.enableProSmartFilesContextMode,
@@ -150,7 +151,7 @@ export async function getModelClient(
               }
             : undefined,
         ),
-        builtinProviderId: "auto",
+        builtinProviderId: model.provider,
       };
 
       return {

src/ipc/utils/llm_engine_provider.ts

@@ -11,6 +11,7 @@ import {
 
 import { OpenAICompatibleChatSettings } from "@ai-sdk/openai-compatible";
 import log from "electron-log";
+import { getExtraProviderOptions } from "./thinking_utils";
 
 const logger = log.scope("llm_engine_provider");
 
@@ -42,6 +43,7 @@ or to provide a custom fetch implementation for e.g. testing.
 */
   fetch?: FetchFunction;
 
+  originalProviderId: string;
   dyadOptions: {
     enableLazyEdits?: boolean;
     enableSmartFilesContext?: boolean;
@@ -113,42 +115,43 @@ export function createDyadEngine(
       defaultObjectGenerationMode:
         "tool" as LanguageModelV1ObjectGenerationMode,
       // Custom fetch implementation that adds files to the request
-      fetch: files?.length
-        ? (input: RequestInfo | URL, init?: RequestInit) => {
-            // Use default fetch if no init or body
-            if (!init || !init.body || typeof init.body !== "string") {
-              return (options.fetch || fetch)(input, init);
-            }
+      fetch: (input: RequestInfo | URL, init?: RequestInit) => {
+        // Use default fetch if no init or body
+        if (!init || !init.body || typeof init.body !== "string") {
+          return (options.fetch || fetch)(input, init);
+        }
 
-            try {
-              // Parse the request body to manipulate it
-              const parsedBody = JSON.parse(init.body);
+        try {
+          // Parse the request body to manipulate it
+          const parsedBody = {
+            ...JSON.parse(init.body),
+            ...getExtraProviderOptions(options.originalProviderId),
+          };
 
-              // Add files to the request if they exist
-              if (files?.length) {
-                parsedBody.dyad_options = {
-                  files,
-                  enable_lazy_edits: options.dyadOptions.enableLazyEdits,
-                  enable_smart_files_context:
-                    options.dyadOptions.enableSmartFilesContext,
-                };
-              }
-
-              // Return modified request with files included
-              const modifiedInit = {
-                ...init,
-                body: JSON.stringify(parsedBody),
-              };
-
-              // Use the provided fetch or default fetch
-              return (options.fetch || fetch)(input, modifiedInit);
-            } catch (e) {
-              logger.error("Error parsing request body", e);
-              // If parsing fails, use original request
-              return (options.fetch || fetch)(input, init);
-            }
+          // Add files to the request if they exist
+          if (files?.length) {
+            parsedBody.dyad_options = {
+              files,
+              enable_lazy_edits: options.dyadOptions.enableLazyEdits,
+              enable_smart_files_context:
+                options.dyadOptions.enableSmartFilesContext,
+            };
           }
-        : options.fetch,
+
+          // Return modified request with files included
+          const modifiedInit = {
+            ...init,
+            body: JSON.stringify(parsedBody),
+          };
+
+          // Use the provided fetch or default fetch
+          return (options.fetch || fetch)(input, modifiedInit);
+        } catch (e) {
+          logger.error("Error parsing request body", e);
+          // If parsing fails, use original request
+          return (options.fetch || fetch)(input, init);
+        }
+      },
     };
 
     return new OpenAICompatibleChatLanguageModel(modelId, restSettings, config);

src/ipc/utils/thinking_utils.ts

@@ -0,0 +1,18 @@
+import { PROVIDERS_THAT_SUPPORT_THINKING } from "../shared/language_model_helpers";
+
+export function getExtraProviderOptions(
+  providerId: string | undefined,
+): Record<string, any> {
+  if (!providerId) {
+    return {};
+  }
+  if (PROVIDERS_THAT_SUPPORT_THINKING.includes(providerId)) {
+    return {
+      thinking: {
+        type: "enabled",
+        include_thoughts: true,
+      },
+    };
+  }
+  return {};
+}