add NAVBACK

imabhijith25 · Jan 24, 2025 · f412768 · f412768
1 parent 841e7c8
commit f412768
Show file tree

Hide file tree

Showing 2 changed files with 57 additions and 33 deletions.
diff --git a/execute.ts b/execute.ts
@@ -14,7 +14,8 @@ export async function runStagehand({
     | "CLOSE"
     | "SCREENSHOT"
     | "OBSERVE"
-    | "WAIT";
+    | "WAIT"
+    | "NAVBACK";
   instruction?: string;
 }) {
   const stagehand = new Stagehand({
@@ -24,38 +25,53 @@ export async function runStagehand({
   await stagehand.init();
 
   const page = stagehand.page;
-  const context = stagehand.context;
 
-  if (method === "GOTO") {
-    await page.goto(instruction!, {
-      waitUntil: "domcontentloaded",
-      timeout: 60000,
-    });
-  }
-  if (method === "ACT") {
-    await page.act(instruction!);
-  }
-  if (method === "EXTRACT") {
-    const { extraction } = await page.extract(instruction!);
-    return extraction;
-  }
-  if (method === "OBSERVE") {
-    return await page.observe({
-      instruction,
-      useAccessibilityTree: true,
-    });
-  }
-  if (method === "CLOSE") {
-    await stagehand.close();
-  }
+  try {
+    switch (method) {
+      case "GOTO":
+        await page.goto(instruction!, {
+          waitUntil: "domcontentloaded",
+          timeout: 60000,
+        });
+        break;
 
-  if (method === "SCREENSHOT") {
-    const cdpSession = await page.context().newCDPSession(page);
-    const { data } = await cdpSession.send("Page.captureScreenshot");
-    return data;
-  }
+      case "ACT":
+        await page.act(instruction!);
+        break;
 
-  if (method === "WAIT") {
-    await new Promise((resolve) => setTimeout(resolve, Number(instruction)));
+      case "EXTRACT": {
+        const { extraction } = await page.extract(instruction!);
+        return extraction;
+      }
+
+      case "OBSERVE":
+        return await page.observe({
+          instruction,
+          useAccessibilityTree: true,
+        });
+
+      case "CLOSE":
+        await stagehand.close();
+        break;
+
+      case "SCREENSHOT": {
+        const cdpSession = await page.context().newCDPSession(page);
+        const { data } = await cdpSession.send("Page.captureScreenshot");
+        return data;
+      }
+
+      case "WAIT":
+        await new Promise((resolve) =>
+          setTimeout(resolve, Number(instruction))
+        );
+        break;
+
+      case "NAVBACK":
+        await page.goBack();
+        break;
+    }
+  } catch (error) {
+    await stagehand.close();
+    throw error;
   }
 }
diff --git a/llm.ts b/llm.ts
@@ -91,13 +91,21 @@ If the goal has been achieved, return "close".`,
         .describe(
           "The reasoning behind the tool call. If the tool is 'CLOSE', this should explain how and why the goal has been achieved."
         ),
-      tool: z.enum(["GOTO", "ACT", "EXTRACT", "OBSERVE", "CLOSE", "WAIT"])
-        .describe(`Tool guidelines:
+      tool: z.enum([
+        "GOTO",
+        "ACT",
+        "EXTRACT",
+        "OBSERVE",
+        "CLOSE",
+        "WAIT",
+        "NAVBACK",
+      ]).describe(`Tool guidelines:
 GOTO: Navigate to a new URL only if not accessible from current page or if you need to navigate to a page to start off
 ACT: Perform a single action on the page
 EXTRACT: Extract data from the page (don't rely on screenshots for text, use the EXTRACT tool and you'll be provided with the result)
 OBSERVE: List available actions when unsure what to do next
 WAIT: Wait for a number of milliseconds
+NAVBACK: Navigate to the previously visited URL
 CLOSE: Close browser when goal is achieved`),
       instruction: z
         .string()