halo-agent 2.6.0 → 2.6.1

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "halo-agent",
-  "version": "2.6.0",
+  "version": "2.6.1",
   "description": "HALO local apply agent — auto-fills job applications using your real Chrome session",
   "main": "index.js",
   "bin": {
@@ -30,7 +30,6 @@
     "vision.js",
     "manusAutomate.js",
     "ui.js",
-    "resolveLinkedIn.js",
     "README.md"
   ],
   "keywords": [

package/poller.js

@@ -31,43 +31,6 @@ async function startPolling(chromeConn, config) {
     const item = await fetchNextJob(config);
     if (!item) continue;
 
-    // ── resolve_linkedin task branch ──────────────────────────────────
-    // Quick (<5s) URL-read tasks live alongside apply tasks. The agent's
-    // Chrome is already authenticated to LinkedIn, so we read the offsite
-    // "Apply on company website" URL from the live DOM and POST it back.
-    // No tab churn, no full apply flow.
-    if (item.intent === 'resolve_linkedin') {
-      active = true;
-      try {
-        const linkedinUrl = item.step_detail || item.apply_url || item.url;
-        ui.info(`${ui.gray('//')} resolving ${ui.bold(item.company)} ${ui.gray('on linkedin')}`);
-        const { resolveLinkedInOnPage } = require('./resolveLinkedIn');
-        const result = await resolveLinkedInOnPage(chromeConn, linkedinUrl).catch((e) => ({
-          resolved_url: null, reason: e?.message || 'resolve threw',
-        }));
-        await fetch(`${config.apiUrl}/apply-queue/${item.id}/resolved`, {
-          method: 'POST',
-          headers: {
-            Authorization: `Bearer ${config.token}`,
-            'Content-Type': 'application/json',
-            ...cfAccessHeaders(config),
-          },
-          body: JSON.stringify({
-            resolved_url: result.resolved_url || null,
-            reason: result.reason || undefined,
-          }),
-        }).catch(() => {});
-        if (result.resolved_url) {
-          ui.info(`  ${ui.sym.tick} ${ui.gray('→')} ${result.resolved_url}`);
-        } else {
-          ui.warn(`  ${ui.sym.cross} could not resolve · ${result.reason || 'no URL found'}`);
-        }
-      } finally {
-        active = false;
-      }
-      continue;
-    }
-
     active = true;
     console.log('');
     ui.info(`${ui.bold(item.company)} ${ui.gray(ui.sym.dot)} ${item.title}`);

package/scanAccessibility.js

@@ -76,6 +76,12 @@ async function injectMmid(page) {
       '[role="option"]',
       '[role="menuitem"]',
       'button[type="submit"]',
+      // ATS toggle-button pairs (Ashby "Yes"/"No", some Greenhouse choice
+      // chips) are rendered as plain `<button>` elements without an explicit
+      // role attribute. Without this, the scan misses every required Yes/No
+      // question on Ashby and the planner silently leaves them blank.
+      // Downstream filters drop nav/close/submit buttons by label heuristic.
+      'button',
       'a[href]',
     ].join(',');
     const all = document.querySelectorAll(sel);
@@ -496,7 +502,7 @@ async function scanFrameDom(frame, frameUrl) {
     const seen = new Set();
     let n = Math.floor(Math.random() * 800000) + 200000; // high offset, no main-frame collision
 
-    const sel = 'input,textarea,select,[contenteditable="true"],[role="textbox"],[role="combobox"],[role="listbox"],[role="radio"],[role="checkbox"],[role="switch"],[role="button"],button[type="submit"]';
+    const sel = 'input,textarea,select,[contenteditable="true"],[role="textbox"],[role="combobox"],[role="listbox"],[role="radio"],[role="checkbox"],[role="switch"],[role="button"],button[type="submit"],button';
     document.querySelectorAll(sel).forEach((el) => {
       const tag = el.tagName.toLowerCase();
       const type = (el.type || '').toLowerCase();
@@ -567,6 +573,7 @@ async function scanFrameDom(frame, frameUrl) {
   return raw.map((f) => ({
     mmid: f.mmid,
     role: f.forceRole || normalizeRole(f),
+    tag: f.tag, // needed by main-frame filter to distinguish <button> from [role=button]
     label: f.label || f.textContent || '',
     description: '',
     required: false,
@@ -651,6 +658,21 @@ async function scanAccessibility(page) {
     if (f.frameUrl) {
       const fl = (f.label || '').trim();
       if (f.role === 'button' && (/^remove\s+\S/i.test(fl) || /^clear\s+selection/i.test(fl) || fl === '×' || fl === '✕' || fl === 'x')) continue;
+      // Apply the same nav-button filter for in-frame buttons.
+      if (f.role === 'button' && f.tag === 'button') {
+        const ll = fl.toLowerCase().trim();
+        const navWords = [
+          'back', 'cancel', 'close', 'menu', 'open menu', 'dismiss',
+          'submit', 'submit application', 'apply', 'apply now',
+          'sign in', 'sign up', 'log in', 'login', 'logout',
+          'next', 'previous', 'continue', 'edit', 'save',
+          'save and continue', 'save & continue',
+          'toggle theme', 'change theme', 'show password', 'hide password',
+        ];
+        if (navWords.includes(ll)) continue;
+        if (f.inputType === 'submit') continue;
+        if (fl.length > 80) continue;
+      }
       if (!fl && !['button', 'link'].includes(f.role)) continue;
       out.push(f); // already in public shape, carries frameUrl
       continue;
@@ -673,6 +695,29 @@ async function scanAccessibility(page) {
     )) {
       continue;
     }
+    // Filter out navigation / submit / close buttons surfaced by the
+    // broader generic-button selector. Answer-choice buttons (Yes / No /
+    // short option text) stay — they're what powers Ashby's toggle pairs.
+    if (normalizeRole(f) === 'button' && f.tag === 'button') {
+      const ll = label.toLowerCase().trim();
+      // Drop common nav buttons by exact-ish label
+      const navWords = [
+        'back', 'cancel', 'close', 'menu', 'open menu', 'dismiss',
+        'submit', 'submit application', 'apply', 'apply now',
+        'sign in', 'sign up', 'log in', 'login', 'logout',
+        'next', 'previous', 'continue',  // multi-step nav — filter; the form's own pagination handles
+        'edit', 'save', 'save and continue', 'save & continue',
+        'toggle theme', 'change theme', 'show password', 'hide password',
+        'view jd', 'view job', 'open application',
+      ];
+      if (navWords.includes(ll)) continue;
+      // Drop submit-type buttons even without a matching label
+      if (f.inputType === 'submit') continue;
+      // Drop very long labels — answer choices are short. >80 chars is
+      // almost certainly help text, fineprint, or accidentally-captured
+      // descendant content.
+      if (label.length > 80) continue;
+    }
     out.push({
       mmid: f.mmid,
       role: normalizeRole(f),

package/resolveLinkedIn.js

@@ -1,161 +0,0 @@
-// resolveLinkedIn — open a LinkedIn job-view URL in the agent's
-// authenticated Chrome session, read the offsite "Apply on company website"
-// URL from the live DOM, return it. The user is already logged into
-// LinkedIn here, so the offsite URL is actually present (unlike the
-// unauthenticated page Cloudflare Browser Run sees).
-//
-// Strategies, in order:
-//   1. <code id="applyUrl"> JSON-encoded URL
-//   2. <a data-tracking-control-name*="apply-link-offsite"> href
-//   3. Voyager / inline JSON dump with companyApplyUrl / applyMethod
-//   4. Click intercept — click the visible Apply button, capture the
-//      destination URL from a new-tab event (works for the modern UI where
-//      the apply button JS-navigates instead of using an <a href>).
-//
-// Returns { resolved_url: string | null, reason?: string }.
-
-const RESOLVE_TIMEOUT_MS = 15_000;
-
-async function resolveLinkedInOnPage(chromeConn, linkedinUrl) {
-  if (!linkedinUrl || !/linkedin\.com\/jobs\/view/i.test(linkedinUrl)) {
-    return { resolved_url: null, reason: 'not a linkedin job-view url' };
-  }
-
-  let page;
-  try {
-    page = await chromeConn.newPage();
-  } catch (e) {
-    return { resolved_url: null, reason: `couldn't open page: ${e.message}` };
-  }
-
-  // Listen for new-tab events on the same context — this catches the
-  // "Apply on company website" button when it opens the ATS in a new tab.
-  let newTabUrl = null;
-  const onNewPage = async (p) => {
-    try {
-      // The new tab navigates through a LinkedIn redirector first, then
-      // lands on the final ATS URL. We wait briefly, then capture.
-      await new Promise((r) => setTimeout(r, 1200));
-      const u = p.url();
-      if (u && /^https?:\/\//i.test(u) && !/linkedin\.com/i.test(u)) {
-        newTabUrl = u;
-      }
-      // Close the new tab — we have what we need, no point loading it.
-      try { await p.close(); } catch {}
-    } catch {}
-  };
-  try { chromeConn.context.on('page', onNewPage); } catch {}
-
-  try {
-    await Promise.race([
-      page.goto(linkedinUrl, { waitUntil: 'domcontentloaded', timeout: RESOLVE_TIMEOUT_MS }),
-      new Promise((_, reject) => setTimeout(() => reject(new Error('navigation timeout')), RESOLVE_TIMEOUT_MS)),
-    ]);
-    // Let LinkedIn's JS render the apply button (lazy-loaded).
-    await page.waitForSelector('#applyUrl, [data-tracking-control-name*="apply"], button[aria-label*="Apply" i], a[aria-label*="Apply" i]', { timeout: 6000 }).catch(() => {});
-
-    // Run the 1-3 strategies inline.
-    const found = await page.evaluate(() => {
-      const isExt = (u) =>
-        !!u && typeof u === 'string' && /^https?:\/\//i.test(u) && !/linkedin\.com/i.test(u);
-
-      // 1. <code id="applyUrl">
-      try {
-        const code = document.getElementById('applyUrl');
-        if (code && code.textContent) {
-          let raw = code.textContent.trim();
-          try { const p = JSON.parse(raw); if (isExt(p)) return p; } catch (e) {}
-          raw = raw.replace(/^"+|"+$/g, '');
-          if (isExt(raw)) return raw;
-        }
-      } catch (e) {}
-
-      // 2. data-tracking-control-name=apply-link-offsite anchors.
-      try {
-        const sel = [
-          'a[data-tracking-control-name*="apply-link-offsite"]',
-          'a[data-tracking-control-name*="apply_unify"]',
-          'a[data-tracking-control-name*="apply-link"]',
-        ].join(', ');
-        const els = Array.from(document.querySelectorAll(sel));
-        for (const el of els) {
-          const href = el.getAttribute('href') || '';
-          if (isExt(href)) return new URL(href, document.baseURI).href;
-        }
-      } catch (e) {}
-
-      // 3. Inline JSON dumps in <code> elements.
-      try {
-        const codes = Array.from(document.querySelectorAll('code'));
-        for (const c of codes) {
-          const txt = c.textContent || '';
-          if (!txt.includes('companyApplyUrl') && !txt.includes('applyMethod')) continue;
-          try {
-            const obj = JSON.parse(txt);
-            const stack = [obj];
-            while (stack.length) {
-              const cur = stack.pop();
-              if (!cur || typeof cur !== 'object') continue;
-              if (isExt(cur.companyApplyUrl)) return cur.companyApplyUrl;
-              if (isExt(cur.applyUrl)) return cur.applyUrl;
-              if (isExt(cur.url) && cur.applyMethod) return cur.url;
-              for (const k of Object.keys(cur)) {
-                if (typeof cur[k] === 'object') stack.push(cur[k]);
-              }
-            }
-          } catch (e) {}
-        }
-      } catch (e) {}
-
-      return null;
-    }).catch(() => null);
-
-    if (found) {
-      // Clean up listener and page.
-      try { chromeConn.context.off('page', onNewPage); } catch {}
-      try { await page.close(); } catch {}
-      return { resolved_url: found };
-    }
-
-    // Strategy 4 — click the Apply button and watch for a new tab.
-    try {
-      const clicked = await page.evaluate(() => {
-        const candidates = [
-          'button[data-tracking-control-name*="apply"]',
-          'a[data-tracking-control-name*="apply"]',
-          'button[aria-label*="Apply" i]',
-          'a[aria-label*="Apply" i]',
-        ];
-        for (const sel of candidates) {
-          const el = document.querySelector(sel);
-          if (el) {
-            try { el.click(); return true; } catch (e) {}
-          }
-        }
-        return false;
-      });
-      if (clicked) {
-        // Wait up to 5s for the new tab listener to populate newTabUrl.
-        const start = Date.now();
-        while (Date.now() - start < 5000 && !newTabUrl) {
-          await new Promise((r) => setTimeout(r, 200));
-        }
-        if (newTabUrl) {
-          try { chromeConn.context.off('page', onNewPage); } catch {}
-          try { await page.close(); } catch {}
-          return { resolved_url: newTabUrl };
-        }
-      }
-    } catch {}
-
-    try { chromeConn.context.off('page', onNewPage); } catch {}
-    try { await page.close(); } catch {}
-    return { resolved_url: null, reason: 'no offsite URL found in any strategy' };
-  } catch (e) {
-    try { chromeConn.context.off('page', onNewPage); } catch {}
-    try { await page.close(); } catch {}
-    return { resolved_url: null, reason: e?.message || 'navigation failed' };
-  }
-}
-
-module.exports = { resolveLinkedInOnPage };