wtagent 0.1.0-alpha.4 → 0.1.0-alpha.5

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wtagent",
-  "version": "0.1.0-alpha.4",
+  "version": "0.1.0-alpha.5",
   "description": "Turn your own web AI session into a local tool-using agent.",
   "repository": {
     "type": "git",

package/src/browser/chatgpt-web-adapter.js

@@ -813,23 +813,33 @@ export class ChatGPTWebAdapter {
   }
 
   async #assistantText(message) {
-    // ChatGPT may render the model's XML inside one of several code blocks,
-    // alongside rich cards (weather, charts, etc.) or other markdown. Search
-    // every code block for the envelope first; if none contains it, fall back
-    // to the full message text so the protocol layer can extract the envelope
-    // from wherever it landed.
+    // Read the whole assistant turn first. A long XML response can contain
+    // Markdown fences inside CDATA; ChatGPT then splits the rendered response
+    // into many <pre><code> nodes and the first node contains the opening tag
+    // but not the closing tag. The parent innerText keeps the complete envelope
+    // and avoids one CDP round trip per nested code block on every poll.
+    const fullText = await message.innerText().catch(() => "");
+    if (fullText.includes("<agent_response")) {
+      return fullText;
+    }
+
+    // Rare fallback for alternate renderers where the parent text omits code
+    // contents: prefer a complete code-block envelope, but retain a partial
+    // one so streaming progress remains visible until its closing tag arrives.
     const codeBlocks = message.locator("pre code");
     const codeBlockCount = await codeBlocks.count();
+    let partialEnvelope = "";
     for (let index = 0; index < codeBlockCount; index += 1) {
       const code = await codeBlocks.nth(index).innerText().catch(() => "");
-      if (code.includes("<agent_response")) {
+      if (hasCompleteAgentEnvelope(code)) {
         return code;
       }
+      if (!partialEnvelope && code.includes("<agent_response")) {
+        partialEnvelope = code;
+      }
     }
-
-    const fullText = await message.innerText().catch(() => "");
-    if (fullText.includes("<agent_response")) {
-      return fullText;
+    if (partialEnvelope) {
+      return partialEnvelope;
     }
 
     const markdown = message.locator(".markdown");

package/src/cli/main.js

@@ -611,7 +611,7 @@ async function runExport(sessionId, options) {
 const program = new Command()
   .name("wtagent")
   .description("Turn your web AI session into a local tool-using agent.")
-  .version("0.1.0-alpha.4")
+  .version("0.1.0-alpha.5")
   .option("--home <path>", "Application data directory")
   .option("--profile-dir <path>", "Dedicated Chrome profile directory")
   .option("--chrome-path <path>", "Chrome/Chromium executable")