surfagent 1.0.7 → 1.0.8

package/API.md

@@ -226,6 +226,26 @@ Get clean, structured readable content from a page. Use this instead of screensh
 
 ---
 
+### POST /dismiss
+
+Dismiss cookie banners, consent dialogs, and modal overlays. Supports 15+ language patterns (English, Norwegian, German, French, Spanish, Italian, Portuguese).
+
+```json
+{ "tab": "0" }
+```
+
+Response:
+```json
+{ "dismissed": [{ "type": "cookie", "text": "reject all" }], "count": 1 }
+```
+
+If nothing was found to dismiss:
+```json
+{ "dismissed": [], "count": 0 }
+```
+
+---
+
 ### POST /focus
 
 Bring a tab to the front in Chrome. Use this when a tab is behind other tabs or windows.
@@ -250,11 +270,18 @@ Click an element on a page.
 { "tab": "0", "selector": "#submit-btn" }
 ```
 
-**By text** (fuzzy match against visible text):
+**By text** (fuzzy match against visible text, including dropdown options):
 ```json
 { "tab": "0", "text": "Submit" }
 ```
 
+**With wait** (wait for page to settle after click — useful for SPA navigation):
+```json
+{ "tab": "0", "text": "Search", "waitAfter": 2000 }
+```
+
+Text search matches: buttons, links, `role="option"`, `role="menuitem"`, `role="listitem"`, `li[aria-label]`, and elements with `onclick`. This means autocomplete dropdown items are clickable by text without needing `/eval`.
+
 **Response:**
 ```json
 { "success": true, "clicked": "BUTTON: Submit" }

package/README.md

@@ -68,7 +68,8 @@ curl -X POST localhost:3456/read -H 'Content-Type: application/json' \
 | `/recon` | POST | Full page map — every element, form, selector, heading, nav link, metadata, captcha detection |
 | `/read` | POST | Structured page content — headings, tables, code blocks, notifications, result areas |
 | `/fill` | POST | Fill form fields with real CDP keystrokes (works with React, Vue, SPAs) |
-| `/click` | POST | Click by CSS selector or text match (handles `target="_blank"` automatically) |
+| `/click` | POST | Click by selector or text, including dropdown options. Optional `waitAfter` for SPAs |
+| `/dismiss` | POST | Auto-dismiss cookie banners, consent dialogs, modals (multi-language) |
 | `/scroll` | POST | Scroll page, returns visible content preview and scroll position |
 | `/navigate` | POST | Go to URL, back, or forward in the same tab |
 | `/eval` | POST | Run JavaScript in any tab or cross-origin iframe |

package/dist/api/act.js

@@ -78,8 +78,11 @@ export async function fillFields(request, options) {
             if (actual === field.value) {
                 results.push({ selector: field.selector, success: true });
             }
+            else if (actual === undefined || actual === null) {
+                results.push({ selector: field.selector, success: false, error: `Element not found or has no value: ${field.selector}` });
+            }
             else {
-                results.push({ selector: field.selector, success: true, error: `Value mismatch: got "${actual}"` });
+                results.push({ selector: field.selector, success: false, error: `Value mismatch: expected "${field.value}", got "${actual}"` });
             }
         }
         catch (error) {
@@ -256,12 +259,14 @@ export async function evalInTab(tab, expression, options) {
     const resolved = await resolveTab(tab, port, host);
     const client = await connectToTab(resolved.id, port, host);
     try {
-        const result = await client.Runtime.evaluate({
+        const timeout = new Promise((_, reject) => setTimeout(() => reject(new Error('Eval timed out after 30s')), 30000));
+        const evalPromise = client.Runtime.evaluate({
             expression,
             returnByValue: true
         });
+        const result = await Promise.race([evalPromise, timeout]);
         await client.close();
-        return result.result.value;
+        return result.result.value ?? null;
     }
     catch (error) {
         await client.close();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "surfagent",
-  "version": "1.0.7",
+  "version": "1.0.8",
   "description": "Browser automation API for AI agents — structured page recon, form filling, clicking, and navigation via Chrome CDP",
   "keywords": [
     "ai-agent",

package/src/api/act.ts

@@ -104,8 +104,10 @@ export async function fillFields(
       const actual = verify.result.value as string;
       if (actual === field.value) {
         results.push({ selector: field.selector, success: true });
+      } else if (actual === undefined || actual === null) {
+        results.push({ selector: field.selector, success: false, error: `Element not found or has no value: ${field.selector}` });
       } else {
-        results.push({ selector: field.selector, success: true, error: `Value mismatch: got "${actual}"` });
+        results.push({ selector: field.selector, success: false, error: `Value mismatch: expected "${field.value}", got "${actual}"` });
       }
     } catch (error) {
       results.push({ selector: field.selector, success: false, error: (error as Error).message });
@@ -329,12 +331,16 @@ export async function evalInTab(
   const client = await connectToTab(resolved.id, port, host);
 
   try {
-    const result = await client.Runtime.evaluate({
+    const timeout = new Promise<never>((_, reject) =>
+      setTimeout(() => reject(new Error('Eval timed out after 30s')), 30000)
+    );
+    const evalPromise = client.Runtime.evaluate({
       expression,
       returnByValue: true
     });
+    const result = await Promise.race([evalPromise, timeout]);
     await client.close();
-    return result.result.value;
+    return result.result.value ?? null;
   } catch (error) {
     await client.close();
     throw error;