halo-agent 2.0.7 → 2.0.9

package/captcha.js

@@ -16,10 +16,43 @@ const CAPSOLVER_API = 'https://api.capsolver.com';
  * Detect if a CAPTCHA is present anywhere on the page (including nested iframes).
  * Returns { detected, type, sitekey, pageUrl }
  */
+// ATS domains that EMBED their form (and its reCAPTCHA) in an iframe on a
+// company-branded careers page. When the captcha lives in one of these
+// frames, the sitekey is registered to THIS domain — so CapSolver must be
+// told the frame URL, not the wrapper page URL. (The Orion case:
+// orioninnovation.com/careers embeds job-boards.greenhouse.io; sending the
+// orioninnovation.com URL → "Invalid domain for site key".)
+const EMBEDDED_ATS_HOSTS = /(greenhouse\.io|lever\.co|ashbyhq\.com|myworkdayjobs\.com|icims\.com|smartrecruiters\.com|jobvite\.com|workable\.com)/i;
+
+/**
+ * Pick the URL to hand CapSolver for a given reCAPTCHA. The sitekey is
+ * domain-locked, so the URL must match the domain the widget actually runs
+ * on. If the form is embedded in a known-ATS iframe, that frame's URL wins
+ * over the top-level page URL.
+ */
+function resolveCaptchaPageUrl(page) {
+  const topUrl = page.url();
+  // If the top URL is itself an ATS domain, use it directly.
+  if (EMBEDDED_ATS_HOSTS.test(topUrl)) return topUrl;
+  // Otherwise, look for an ATS iframe — its URL is where the sitekey lives.
+  for (const frame of page.frames()) {
+    if (frame === page.mainFrame()) continue;
+    const fUrl = frame.url();
+    if (fUrl && EMBEDDED_ATS_HOSTS.test(fUrl)) {
+      // Strip the recaptcha frame itself — we want the FORM's frame, not
+      // google's. recaptcha frames are on google.com so they won't match
+      // EMBEDDED_ATS_HOSTS anyway, but be defensive.
+      if (!/google\.com|gstatic\.com/i.test(fUrl)) return fUrl;
+    }
+  }
+  return topUrl;
+}
+
 async function detectCaptcha(page) {
-  // The pageUrl CapSolver wants is the top-level URL the user sees, not an
-  // inner iframe URL. Anchor it here so every branch returns the same thing.
-  const pageUrl = page.url();
+  // The URL CapSolver needs is the domain the sitekey is registered to —
+  // the embedded ATS frame's URL when the form is iframed into a branded
+  // careers page, otherwise the top-level URL.
+  const pageUrl = resolveCaptchaPageUrl(page);
 
   // First try the top-level frame via evaluate
   const topResult = await page.evaluate(() => {

package/orchestrator.js

@@ -87,6 +87,11 @@ async function runJob(queueItem, chromeConn, config, reportStatus) {
 
   let page = null;
   let tempResumeFile = null;
+  // When the run ends in a terminal SUCCESS (DONE), we leave the tab open
+  // so the user sees the confirmation page — visual loop-closure. Only
+  // close the tab on errors/cleanup. Set true right before each successful
+  // DONE return/fall-through.
+  let leaveTabOpen = false;
 
   // Initialize cross-page context tracker
   const ctx = createFormContext();
@@ -246,6 +251,7 @@ async function runJob(queueItem, chromeConn, config, reportStatus) {
         const currentUrl = page.url();
         const looksLikeConfirmation = /thank|confirm|success|applied|submitted/i.test(currentUrl);
         if (looksLikeConfirmation) {
+          leaveTabOpen = true;
           await reportStatus('DONE', { fields_filled: cumulativeFilled });
           await clearCheckpoint(config, queueId);
           console.log(`[orchestrator] Detected confirmation page by URL: ${currentUrl}`);
@@ -388,6 +394,7 @@ async function runJob(queueItem, chromeConn, config, reportStatus) {
           if (vRes.ok) vVerdict = await vRes.json();
         } catch {}
         if (vVerdict.submitted === true) {
+          leaveTabOpen = true;
           await reportStatus('DONE', { confirmation_screenshot_r2_key: confirmKey || null, fields_filled: cumulativeFilled });
           await clearCheckpoint(config, queueId);
           console.log(`[orchestrator] Done via vision (verified): ${queueItem.company} - ${queueItem.title}`);
@@ -613,6 +620,7 @@ async function runJob(queueItem, chromeConn, config, reportStatus) {
         fields_filled: cumulativeFilled,
       });
     } else {
+      leaveTabOpen = true;
       await reportStatus('DONE', {
         confirmation_screenshot_r2_key: confirmKey || null,
         fields_filled: cumulativeFilled,
@@ -668,9 +676,16 @@ async function runJob(queueItem, chromeConn, config, reportStatus) {
     if (tempResumeFile) {
       try { fs.unlinkSync(tempResumeFile); } catch {}
     }
-    // Close the tab (leave other tabs untouched)
-    if (page) {
+    // Close the tab on failure/cleanup — but on a successful submit, LEAVE
+    // it open so the user sees the confirmation page (visual loop-closure).
+    // The next job opens its own new tab, so leaving this one doesn't block
+    // anything; sequential processing is enforced by the poller (one job at
+    // a time). Stale success tabs accumulate, but that's the user's call to
+    // close — they asked for the satisfaction of seeing it land.
+    if (page && !leaveTabOpen) {
       try { await page.close(); } catch {}
+    } else if (page && leaveTabOpen) {
+      console.log('[orchestrator] Leaving confirmation tab open for your review.');
     }
   }
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "halo-agent",
-  "version": "2.0.7",
+  "version": "2.0.9",
   "description": "HALO local apply agent — auto-fills job applications using your real Chrome session",
   "main": "index.js",
   "bin": {