agentleFS
carsanmilcarcarsanmilcar/job-search-template/.claude/skills/linkedin-job-search/SKILL.md

linkedin-job-search

Skill1 starsChanged 54 days agoMITView on GitHub

Search and read LinkedIn job listings in the user's Chrome using targeted JS extraction instead of full-page snapshots. Works with Claude in Chrome (preferred) or the Chrome DevTools MCP. Use whenever the user asks to search/browse LinkedIn for job offers, or to pull details from a specific LinkedIn job posting URL.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. MIt9/job-search-skills★ 0same name
    2. rubenviolinha/linkedin-job-search-skill★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repositoryOne other file it gives its agents.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Check your tool list (load deferred tools with ToolSearch if needed) and use the first that exists:

    line 15

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

linkedin-job-search

LinkedIn pages are enormous (nav, ads, "gente que podrías conocer", footer with 30 language options...). Screenshots, accessibility snapshots or read_page on a LinkedIn page burn thousands of tokens on noise. This skill runs a small JS snippet in the page to pull only the fields needed, straight from the DOM.

Step 0 — which browser tool is available?

Check your tool list (load deferred tools with ToolSearch if needed) and use the first that exists:

OptionNavigateRun JSHow to pass the snippets below
A. Claude in Chrome (preferred)mcp__claude-in-chrome__navigatemcp__claude-in-chrome__javascript_tool (action: "javascript_exec", tabId, text)Paste the snippet as-is in text (they are IIFEs; the tool returns the value of the last expression and supports top-level await). Get a tabId first with tabs_context_mcp.
B. Chrome DevTools MCPmcp__chrome-devtools__navigate_pagemcp__chrome-devtools__evaluate_scriptPass only the arrow function in function — drop the leading await (, the wrapping ( and the final )(); the tool calls it itself. Use list_pages/select_page to reuse an open LinkedIn tab.

If neither exists, don't improvise with WebFetch (LinkedIn blocks it). Tell the user, in plain language, that the browser connection isn't set up and point them to the README section "Conectar Claude con Chrome". Meanwhile they can paste the offer text or URL by hand, which is enough for everything else in this repo.

Login: the user must be logged into LinkedIn in that Chrome. If the page shows a login wall, ask them to log in themselves — never type credentials.

Searching job listings

Build the search URL directly instead of clicking through the UI:

https://www.linkedin.com/jobs/search/?keywords=<url-encoded keywords>&location=<url-encoded location>&f_TPR=<r604800|r2592000>
  • f_TPR=r604800 → last 7 days, r2592000 → last 30 days. Omit for no date filter.
  • f_WT=2 → remote only (1 on-site, 3 hybrid; combine as f_WT=2%2C3).
  • Boolean OR in keywords is unreliable — LinkedIn sometimes collapses results. Prefer one keyword per search and run several.
  • Navigation may report a timeout even when the page loaded (LinkedIn keeps background requests open) — run the extraction anyway and check the result.

LinkedIn only renders the cards you have scrolled to (~7 of 25 at first). Scroll the results list first with this snippet (it finds the scrollable pane itself):

await (async () => {
  const first = document.querySelector('li[data-occludable-job-id], div.job-card-container');
  let pane = first && first.parentElement;
  while (pane && pane !== document.body && !(pane.scrollHeight > pane.clientHeight + 50 && getComputedStyle(pane).overflowY !== 'visible')) pane = pane.parentElement;
  const target = (pane && pane !== document.body) ? pane : document.scrollingElement;
  for (let y = 0; y <= target.scrollHeight; y += 400) { target.scrollTop = y; await new Promise(r => setTimeout(r, 250)); }
  return new Set([...document.querySelectorAll('a[href*="/jobs/view/"]')].map(a => a.href.split('?')[0])).size;
})()

Then extract job cards (covers both the logged-in and the logged-out page; dedupes by URL because LinkedIn often renders each card twice):

(() => {
  const results = [];
  document.querySelectorAll('div.job-card-container, li.jobs-search-results__list-item, li[data-occludable-job-id], div.base-card').forEach(card => {
    const title = card.querySelector('a.job-card-list__title, .job-card-list__title, .base-search-card__title, a[href*="/jobs/view/"]');
    const company = card.querySelector('.job-card-container__primary-description, .artdeco-entity-lockup__subtitle, .job-card-container__company-name, .base-search-card__subtitle');
    const location = card.querySelector('.job-card-container__metadata-item, .artdeco-entity-lockup__caption, .job-search-card__location');
    const link = card.querySelector('a[href*="/jobs/view/"]');
    if (title || link) {
      results.push({
        title: title ? title.innerText.trim() : null,
        company: company ? company.innerText.trim() : null,
        location: location ? location.innerText.trim() : null,
        url: link ? link.href.split('?')[0] : null
      });
    }
  });
  const seen = new Set();
  const dedup = results.filter(r => { if (seen.has(r.url)) return false; seen.add(r.url); return true; });
  return JSON.stringify({count: dedup.length, results: dedup});
})()

LinkedIn loads ~25 cards per page; add &start=25, &start=50... for more. If a search returns 0 results it usually means there are genuinely no matches — only suspect the selectors if every search returns 0 (then fall back to get_page_text / reading main.innerText).

Reading a single job posting

Navigate to https://www.linkedin.com/jobs/view/<id>/, then extract the text (not the page):

(() => {
  const main = document.querySelector('main');
  const text = main ? main.innerText : document.body.innerText;
  return JSON.stringify({text: text.slice(0, 4000)});
})()

4000 chars covers title, company, "Acerca del empleo", requirements and benefits for almost every posting. Increase the slice only if requirements are cut off at the tail.

Workflow in this repo

  1. Show the user a short list (title, company, location, why it fits or not according to kb/profile.md → ## Criterios de búsqueda). Don't dump every card.
  2. Log the ones worth tracking in applications.md.
  3. For an offer they want to pursue, save the literal text + URL in applications/<empresa-rol>/oferta.md (postings disappear), and put % Oferta: <url> near the top of the CV's main.tex.
  4. Close the browser tabs you opened when done.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.