feat: universal JD extraction + friendlier reload error message

Compile confirmed OK via build log marker TECTONIC_REAL_WARMUP_OK. This
change touches only JD extraction and error messaging; the ATS pipeline,
background worker, and PDF download flow are untouched.

content.js: set scoped when a real description selector matched or when
JD-signal phrases are present, instead of depending on a fragile detail
pane container. Fixes LinkedIn collections/easy-apply and other variants.
Add JD_SIGNALS and _hasJdSignal so any platform whose text reads like a
JD is trusted and bypasses the listing-junk guard. Broaden generic
selectors to Workday, Lever, Greenhouse, Ashby and company sites. Bump
the lazy-load wait to 2.5s and add a one-time retry for slow SPA pages.

popup.js: detect the post-update context-invalidated error and tell the
user to refresh the page and run again, instead of the raw error.

manifest 1.6.1 to 1.7.0

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

HISTORY.md

@@ -4,6 +4,40 @@ A running log of everything built, fixed, and changed. Most recent first.
 
 ---
 
+## 2026-06-24 (AM1) — Compile confirmed OK; universal JD extraction + friendlier reload error
+
+Build log for 29baf71 shows `TECTONIC_REAL_WARMUP_OK` (29.75 KiB PDF) — the
+PDF-compile saga is resolved end-to-end. Remaining work was extension JD
+extraction; the ATS pipeline, background worker, and PDF/download flow are left
+untouched per request.
+
+**Universal, robust JD extraction (`extension/content.js`):**
+- `scoped` is now set whenever a real description selector matched (length ≥ 80)
+  OR the text carries JD-signal phrases — no longer dependent on a fragile
+  `detailPane` container match. Fixes LinkedIn `/collections/easy-apply/` and the
+  other LinkedIn URL variants that previously reported "not a job posting".
+- New `JD_SIGNALS` + `_hasJdSignal()`: any platform whose extracted text reads
+  like a JD ("responsibilities", "about the job", "what you'll do", …) is trusted
+  and bypasses the listing-junk guard.
+- `extractGenericJD` candidate set broadened to the major ATS/company platforms:
+  Workday (`[data-automation-id=jobPostingDescription]`), Lever
+  (`[data-qa=job-description]`), Greenhouse (`.job__description`), Ashby,
+  SmartRecruiters, plus `[class*=description]` — so it works on company career
+  sites, not just LinkedIn.
+- SPA timing: lazy-load wait 1.2s → 2.5s, plus a ONE-TIME retry (900ms) in the
+  EXTRACT_JD handler when the first pass is weak — covers late-hydrating detail
+  panes on LinkedIn/Indeed/company SPAs.
+
+**Friendlier "Extension context invalidated" (`extension/popup/popup.js`):**
+That error only happens when the extension is updated while a page keeps the OLD
+content script. handleResult now detects it (and "message port closed"/"receiving
+end does not exist") and shows "The extension was just updated. Please refresh this
+page (F5), then click Run again." instead of the raw error.
+
+manifest 1.6.1 → 1.7.0.
+
+---
+
 ## 2026-06-23 (PM14) — Fix Tectonic segfault (sanitize fontawesome OTF) + dedupe keywords
 
 Build log (commit 65ea09d) gave the smoking gun: Tectonic SEGFAULTS loading

extension/content.js

@@ -113,11 +113,23 @@ function scrubOverlayLines(text) {
  * @returns {string}
  */
 function extractGenericJD() {
-  // Candidate containers — ordered from most-specific to least-specific
+  // Candidate containers — ordered from most-specific to least-specific. Covers
+  // the major ATS/company-site platforms so the extension works everywhere, not
+  // just LinkedIn: Greenhouse, Lever, Ashby, Workday, SmartRecruiters, etc.
   const candidates = [
     ...document.querySelectorAll(
       'article, main, [role="main"], .description, .job-description, ' +
-      '.job-details, .posting-description, #job-details, #jobDescriptionText'
+      '.job-details, .posting-description, #job-details, #jobDescriptionText, ' +
+      // Workday
+      '[data-automation-id="jobPostingDescription"], ' +
+      // Lever
+      '[data-qa="job-description"], .section-wrapper.page-full-width, ' +
+      // Greenhouse
+      '#content .job__description, .job__description, #app_body, ' +
+      // Ashby / SmartRecruiters / Recruitee / generic
+      '.ashby-job-posting-right-pane, .job-sections, .jobAd, ' +
+      '[class*="jobDescription"], [class*="job-description"], [class*="JobDescription"], ' +
+      '[class*="posting"], [class*="description"]'
     )
   ];
 
@@ -205,8 +217,9 @@ function parseDocTitle() {
 async function fastJD(selectors, root) {
   const now = firstMatch(selectors, root);
   if (now.length > 50) return now;
-  // Single short wait on the most-likely selector for SPA lazy-load.
-  const el = await waitForElement(selectors[0], 1200, root);
+  // Wait on the most-likely selector for SPA lazy-load (description panes on
+  // LinkedIn/Indeed hydrate after navigation). 2.5s keeps it fast but reliable.
+  const el = await waitForElement(selectors[0], 2500, root);
   if (el) {
     const txt = (el.innerText || el.textContent || '').trim();
     if (txt.length > 50) return txt;
@@ -273,16 +286,20 @@ async function extractLinkedIn() {
     '.description__text',
   ], scope);
 
-  // If selectors missed, fall back to the densest text block WITHIN the detail
+  // A description selector matched (`#job-details` is the stable anchor across
+  // ALL LinkedIn variants: /jobs/view, /collections, /collections/easy-apply,
+  // /search). That text is the posting, never the left list → trust it directly.
+  let scoped = !!(jd_text && jd_text.length >= 80);
+
+  // Selectors missed → scope the densest-block fallback to the right-hand detail
   // pane (never the whole page, so the left job list can't win the contest).
-  let scoped = false;
-  if ((!jd_text || jd_text.length < 50) && detailPane) {
-    jd_text = scrubOverlayLines(stripChrome(detailPane)).trim();
-  }
-  if (jd_text && jd_text.length >= 80 && detailPane) {
-    // We extracted from a real detail pane → trust it; skip the listing-junk
-    // heuristic (a detail pane legitimately contains an "Easy Apply" button).
-    scoped = true;
+  if (!scoped) {
+    if ((!jd_text || jd_text.length < 50) && detailPane) {
+      jd_text = scrubOverlayLines(stripChrome(detailPane)).trim();
+    }
+    if (jd_text && jd_text.length >= 120 && _hasJdSignal(jd_text)) {
+      scoped = true;   // detail-pane text that reads like a JD
+    }
   }
   if (!jd_text || jd_text.length < 50) {
     jd_text = extractGenericJD();
@@ -363,6 +380,12 @@ async function extractJD() {
     result.job_title = result.job_title || dt.job_title;
     result.company = result.company || dt.company;
   }
+  // Any platform: if the extracted text reads like a real JD, mark it trusted so
+  // the listing-junk guard never blocks it (company sites, Greenhouse, Lever, …).
+  if (!result.scoped && _hasJdSignal(result.jd_text) &&
+      (result.jd_text || '').length >= 200) {
+    result.scoped = true;
+  }
   return result;
 }
 
@@ -480,6 +503,23 @@ const NON_JOB_TITLES = [
   'collections', 'my jobs', 'saved jobs', 'recommended jobs', 'jobs for you',
 ];
 
+// Phrases that reliably indicate the text IS a job description (any platform).
+// Used to accept a JD even on split-view pages where some "Easy Apply" chrome
+// leaks in, so the listing-junk guard never blocks a real posting.
+const JD_SIGNALS = [
+  'responsibilit', 'requirement', 'qualification', 'about the job',
+  'about the role', 'about the company', "what you'll do", 'what you will do',
+  'who you are', 'you will', 'we are looking', 'role overview', 'job description',
+  'preferred', 'minimum qualifications', 'key skills', 'what we', 'your role',
+  'in this role', 'nice to have', 'must have', 'day to day', 'day-to-day',
+];
+
+/** True if `text` contains language characteristic of a real job description. */
+function _hasJdSignal(text) {
+  const l = (text || '').toLowerCase();
+  return JD_SIGNALS.some((s) => l.includes(s));
+}
+
 /**
  * True when the extracted result is a jobs LIST / search / recommendations page
  * rather than a single job posting. Such pages yield junk "keywords" (e.g.
@@ -517,24 +557,33 @@ chrome.runtime.onMessage.addListener((msg, sender, sendResponse) => {
   }
 
   if (msg.type === 'EXTRACT_JD') {
-    extractJD()
+    // Extract, and if the page clearly hadn't loaded the posting yet (short text
+    // and no JD signal), retry ONCE after a short delay — covers SPA navigations
+    // on LinkedIn/Indeed/company sites where the detail pane hydrates late.
+    const tryExtract = () => extractJD();
+    const finalize = (result) => {
+      if (!result.scoped && looksLikeListingJunk(result)) {
+        result.extraction_failed = true;
+        result.extraction_reason = 'not_a_job_posting';
+        result.jd_text = '';
+      } else if (!result.jd_text || result.jd_text.length < 80) {
+        result.extraction_failed = true;
+        result.jd_text = '';
+      }
+      if (result.jd_text && result.jd_text.length > 16000) {
+        result.jd_text = result.jd_text.slice(0, 16000);
+      }
+      sendResponse(result);
+    };
+    tryExtract()
       .then((result) => {
-        // Reject jobs LIST / search / recommendation pages — they have no single
-        // JD and produce junk keywords. Surface a clear reason so the popup tells
-        // the user to open a specific posting (or paste the JD) instead of
-        // silently generating a 0%-coverage resume from page chrome.
-        if (!result.scoped && looksLikeListingJunk(result)) {
-          result.extraction_failed = true;
-          result.extraction_reason = 'not_a_job_posting';
-          result.jd_text = '';
-        } else if (!result.jd_text || result.jd_text.length < 80) {
-          result.extraction_failed = true;
-          result.jd_text = '';
-        }
-        if (result.jd_text && result.jd_text.length > 16000) {
-          result.jd_text = result.jd_text.slice(0, 16000);
+        const weak = (!result.jd_text || result.jd_text.length < 80)
+          && !looksLikeListingJunk(result);
+        if (weak) {
+          setTimeout(() => tryExtract().then(finalize).catch(() => finalize(result)), 900);
+        } else {
+          finalize(result);
         }
-        sendResponse(result);
       })
       .catch((err) => {
         sendResponse({

extension/manifest.json

@@ -1,7 +1,7 @@
 {
   "manifest_version": 3,
   "name": "ATS Resume Generator",
-  "version": "1.6.1",
+  "version": "1.7.0",
   "description": "Tailors your resume (PDF or LaTeX) to any job posting using the ATS pipeline.",
   "permissions": [
     "storage",

extension/popup/popup.js

@@ -448,7 +448,15 @@ function handleResult(result) {
     statusEl.innerHTML = '';
     const line = document.createElement('div');
     line.style.color = 'red';
-    line.textContent = `Error: ${msg}`;
+    // "Extension context invalidated" happens when the extension was just
+    // updated/reloaded while this page kept the OLD content script. A page
+    // refresh re-injects the new one. Give that exact instruction instead of
+    // the raw error.
+    if (/context invalidated|message port closed|receiving end does not exist/i.test(msg)) {
+      line.textContent = 'The extension was just updated. Please refresh this page (F5), then click Run again.';
+    } else {
+      line.textContent = `Error: ${msg}`;
+    }
     statusEl.appendChild(line);
     // Config errors → offer a one-click jump to Options.
     if (/option|api url|token|not_configured|configure/i.test(msg)) {