Claude Job Search: A Daily Digest of New Jobs Ranked Against Your CV
Create a Claude Cowork scheduled task that runs every morning. It calls two Apify job scrapers (LinkedIn and Indeed) for postings from the last 24 hours, scores each one against the CV you pasted into the task, and gives you the top 5 with tailoring notes. The notes can only reuse lines from your CV, so Claude cannot invent experience for you. At the job volumes below, the scraping stays inside Apify's free $5 monthly credit.
Job alerts from the boards themselves match on keywords, so you get 40 emails a week and skim them all. The job that matters is the reverse: read every new posting, compare it to what you have actually done, and surface the few worth an application today. That is reading work, which is what Claude is good at, provided something fetches the postings for it.
What you end up with
Each morning a new Cowork session is waiting with a short report:
- The top 5 new postings, from LinkedIn and Indeed, deduplicated across both boards.
- A fit score out of 100 for each, with one line on why.
- Gaps: the requirements your CV does not show, stated plainly.
- 3 to 5 tailored bullets per job, each one rewritten from a specific line of your CV, ready to drop into the application.
- A "maybe" list of the next 10, title and link only, so nothing good disappears below the cut.
You still decide what to apply for, and you send every application yourself. The digest replaces the skimming, not the judgement.
Let Claude set it up for you
Paste this into a new Cowork task. Claude asks for your search details one at a time, checks the Apify connection, runs a small priced test, then builds the daily task with you. The manual walkthrough below explains every piece if you would rather do it by hand.
Claude interviews you, tests the two job scrapers on a tiny sample, and then creates the daily scheduled task, asking before anything that costs money.
- Asks for your job title, location, filters, preferences and CV, one question at a time
- Checks the Apify connector is connected, and explains how to add it if not
- Runs a 5-job test on each board and shows the cost
- Writes the final task prompt and sets up the daily schedule with you
Help me set up a daily job search digest in Cowork, using the Apify connector.
Work through these stages in order, and ask me ONE question at a time.
1. INTERVIEW. Ask me, one at a time:
- the job title(s) I am targeting
- the city and country (and whether remote roles count)
- hard filters: languages I lack, minimum salary, agencies to exclude
- my preferences (company size, industry, remote/hybrid), in order of importance
- daily or weekdays only
- my full master CV as plain text (tell me to paste it, not attach a file,
so the scheduled task can run while my computer is off)
2. CHECK THE CONNECTOR. Confirm you can use Apify tools (for example call-actor
and get-dataset-items, or the Actors listed as their own tools). If you cannot,
stop and tell me how to add a custom connector in Cowork (Customize > Connectors
> Add custom connector) with this URL, then wait for me:
https://mcp.apify.com?tools=actors,curious_coder/linkedin-jobs-scraper,valig/indeed-jobs-scraper
3. TEST SMALL. Before running anything, tell me what you will run and ask me to
confirm. Then run:
- curious_coder/linkedin-jobs-scraper with my keywords and location,
datePosted "past24Hours", limitPerSource 5
- valig/indeed-jobs-scraper with my title, location and country code,
datePosted "1", limit 5
Show me the jobs you got, the score you would give each, and what the test
cost according to the run details. Ask whether the results look right.
4. BUILD THE TASK. Write the final scheduled-task prompt: search both boards
for the last 24 hours (limit 40 each), merge duplicates, keep at most 2
postings per company, apply my filters,
score each job out of 100 against my CV (40 must-haves, 25 seniority, 20 domain,
15 my preferences), report the top 5 with link, score, one-line reason, Gaps,
and 3-5 tailored bullets, then a Maybe list of the next 10.
Bullet rules: every bullet must rewrite a specific CV line and quote that line
in [brackets]; never add any skill, tool, metric, employer or result that is not
in my CV; unmatched requirements go under Gaps. Never pad the list.
Estimate the maximum monthly scraping cost at those limits.
Show me the full prompt and ask me to confirm before you create the schedule.
If you cannot create a scheduled task from this conversation, give me the
prompt and tell me exactly where to paste it (Scheduled > New task).
Never apply to jobs or contact anyone on my behalf.
Claude asks before each change it makes. Read each step before you approve it.
Paste into a normal Cowork task with a week of digest reports pasted below it.
Below are this week's job digest reports and my task prompt. List the jobs I
would most likely skip that scored above 70, and the jobs in the Maybe lists that look like strong fits. Suggest
specific changes to the filters and scoring weights in my task prompt, and show
me the revised prompt. Do not change the bullet rules.
[paste the week's reports and your current task prompt here]
What you need
| Requirement | Why |
|---|---|
| A paid Claude plan (Pro, Max, Team or Enterprise) | Scheduled tasks are a Cowork feature on all paid plans. |
| An Apify account | Runs the job-board scrapers. The free plan includes $5 of usage credit per month. |
| Apify connected to Cowork | A one-time setup, covered in connecting Apify to Claude Cowork. |
| Your master CV as plain text | Everything you have done, longer than any CV you would send. Claude tailors down from it, never up. |
Step 1: Load only the two job scrapers
Apify's MCP server can expose its whole Store, but a daily task only needs two tools. Naming them in the connector URL keeps Claude from browsing for alternatives or starting other paid Actors on every run. The context saving is small: Claude loads tool definitions on demand, so the limit is about control and cost, not tokens (we measured it). If you already added Apify as a custom connector, edit its URL to:
https://mcp.apify.com?tools=curious_coder/linkedin-jobs-scraper,valig/indeed-jobs-scraper
The two Actors, as of October 2026:
| Actor | Inputs this task uses | Price per job on the free plan |
|---|---|---|
LinkedIn Jobs Scraper (curious_coder) | keywords, location, datePosted: past24Hours, limitPerSource | $0.002 subject to change · verified 2026-10-21 |
Indeed Jobs Scraper (valig) | title, location, country, datePosted: "1" (last 24 hours), limit | $0.0001 subject to change · verified 2026-10-21 |
Both charge less per job on paid Apify plans. Want a different board, such as Glassdoor or a remote-only site? The Apify jobs category roundup lists the alternatives; add the Actor ID to the same tools list.
The "last 24 hours" filter is what makes the digest daily. Each run only sees postings published since the previous run, so the task needs no memory of what it showed you yesterday.
What a real test run returned
We ran both Actors on 21 October 2026 with a limit of 5 each: "Senior Product Designer" in Berlin on LinkedIn, and "Product Designer" in Berlin on Indeed, both restricted to the last 24 hours. Real output from a test run on 2026-10-21, job titles only:
LinkedIn (keywords: Senior Product Designer) | Indeed (title: Product Designer) |
|---|---|
| Senior Product Designer (w/m/d/x), with German at C1 level | Head of Funnel Conversion (CRO), remote |
| Product Designer (AI x Greentech) (m/f/d) | Product Growth Lead (Referrals & Viral Loops), remote |
| E-Commerce UX & Content Manager (m/w/d) | Funnel Conversion (CRO) Lead, remote |
| Digital-Konzepter:in (UI/UX, Strategie) | Web Designer, remote |
| Kommunikationsdesigner:in (Webflow) | Product Manager, Engagement, remote |
Three things this shows, and the prompt below handles each:
- The date filter works. Every posting on both boards was published on 20 October, inside the 24 hours before the run. The LinkedIn Actor's log shows it turned the filter into LinkedIn's own "past 24 hours" search parameter.
- Keyword matching is loose. Only 2 of the 5 LinkedIn results were product design roles, and one of those required German at C1 level. The search fetches candidates; the scoring and filters do the real selection, which is why the hard filters exist.
- One employer can flood a board. All 5 Indeed results came from a single company posting near-identical remote listings, and only one was a design role. The prompt caps each company at 2 postings.
The test cost $0.0051 for 5 LinkedIn jobs and $0.0014 for 5 Indeed jobs, start fees included, on a paid Apify plan. Free-plan rates are slightly higher; the cost section below uses them.
Step 2: Write the task prompt
This prompt is the whole product. Paste it into the task and replace the bracketed parts. Your CV goes inside the prompt, not as a file on your computer: Anthropic notes that a scheduled task needing local files or apps only runs locally, and you want this one to run while your laptop is closed.
Every run, do the following.
SEARCH
1. Run curious_coder/linkedin-jobs-scraper with keywords "[Senior Product Designer]",
location "[Berlin, Germany]", datePosted "past24Hours", limitPerSource 40.
2. Run valig/indeed-jobs-scraper with title "[Senior Product Designer]",
location "[Berlin]", country "[de]", datePosted "1", limit 40.
3. Merge the results. Treat two postings with the same company and a near-identical
title as one job, and keep both links.
FILTER (drop silently)
- Anything requiring [German at C1 level].
- Agencies or recruiters that do not name the hiring company.
- More than 2 postings from the same company: keep only the 2 that score highest.
- Posted salary below [€70,000] when a salary is stated. Keep postings with no salary.
SCORE each remaining job out of 100 against MY CV below:
- 40 points: must-have requirements I can show from the CV.
- 25 points: seniority and scope match (team size, ownership, budget).
- 20 points: domain match (industry, product type, users).
- 15 points: my preferences: [remote or hybrid, B2B SaaS, team under 200 people].
REPORT
- The top 5 by score. For each: title, company, link(s), score, one-line reason,
"Gaps:" (requirements the CV does not show), and 3 to 5 tailored bullets.
- Then a "Maybe" list: the next 10, title, company and link only.
- If fewer than 5 jobs pass the filter, say so. Do not pad the list.
HARD RULES FOR BULLETS
- Every bullet must be a rewrite of a specific line in my CV. After each bullet,
quote the CV line it came from in [brackets].
- Never add a skill, tool, metric, employer, title or result that is not in the CV.
- If a requirement has no matching CV line, it goes under Gaps, never into a bullet.
MY CV
[paste your full master CV here]
Three choices in this prompt matter more than the rest.
The bracketed CV quote is the anti-hallucination check. Asking a model "don't make things up" is weaker than asking it to show its source for every claim. When each bullet has to carry the CV line it came from, an invented bullet has nothing to quote, and you can spot a stretched one at a glance.
Gaps are a section, not an omission. A digest that only tells you why you fit trains you to apply for jobs you will be screened out of. Listing what is missing lets you skip those, or address the gap in a cover letter honestly.
"Do not pad the list." On a quiet day there may be two good jobs. Without this line, the model fills five slots anyway.
Step 3: Schedule it
In the Claude desktop app, open Scheduled in the left sidebar, choose New task, and either let Claude set it up from a description or set it up manually with the prompt above. Pick daily, or weekdays if recruiters in your market do not post at weekends.
Each run happens as its own Cowork session, so the morning's report is in that session's history. Cowork's scheduled tasks are covered in more depth in web-data digests on autopilot.
Before you trust the schedule, paste the same prompt into an ordinary Cowork task, run it once, and read the result critically. Two things to check:
- Are the filters doing their job? If agency postings get through, name the agencies in the filter.
- Do the scores match your gut? If a job you would never take scores 85, the preferences weight is too low or too vague. Rewrite it in concrete terms.
What it costs
The scheduling is part of your Claude plan. The scraping is billed by Apify per job returned, plus a small start fee per run: $0.001 for the Indeed Actor and $0.00005 for the LinkedIn Actor on the free plan.
At the limits in the prompt, a busy day returns up to 40 LinkedIn and 40 Indeed postings:
| Per day (max) | Per 30 days (max) | |
|---|---|---|
| LinkedIn, 40 × $0.002 | $0.08 | $2.40 |
| Indeed, 40 × $0.0001 | $0.004 | $0.12 |
| Start fees, 1 run each | $0.001 | $0.03 |
| Total | about $0.09 | about $2.55 |
That fits inside Apify's $5 monthly free credit, with room to add a second search (another city, or an adjacent title). Most days return fewer than the limit, so the real figure is usually lower. Prices are the free-plan rates checked on 21 October 2026; check each Actor's page before scaling up, and see the Apify pricing guide for how credits work.
Limits worth knowing
- LinkedIn's terms restrict automated access. The scraper reads public job listings, not your account, but read the legal and policy notes on LinkedIn scraping before you depend on it. Indeed alone is a reasonable fallback.
- Scores are an opinion, not a filter you can audit. Once a week, skim the Maybe list. If good jobs keep landing there, adjust the weights.
- Salaries are only as good as the posting. Many ads still omit pay, which is why the prompt keeps postings with no salary rather than dropping them.
- Never let it apply for you. Keep the task to reading and drafting. Sending applications on your behalf is exactly the step where an invented detail would cost you.
FAQ
Yes. A Claude Cowork scheduled task can run daily, call Apify's LinkedIn and Indeed job scrapers through the Apify connector, filter postings from the last 24 hours, and score them against a CV you paste into the task. Scheduled tasks are available on all paid Claude plans.
It can, if you only ask it not to. The more reliable fix is to require every tailored bullet to quote the CV line it was rewritten from, and to put any unmatched requirement in a separate Gaps section. An invented bullet then has no source to quote, which makes it easy to spot.
Yes, as long as the task does not need local files or apps. Cowork scheduled tasks run remotely, so paste your CV into the task prompt rather than pointing it at a file on your computer, and use Apify's remote MCP server as the connector.
The schedule is included in paid Claude plans. Scraping is billed by Apify per job: on the free plan, about $0.002 per LinkedIn posting and $0.0001 per Indeed posting as of October 2026. A daily run capped at 40 postings per board costs at most about $2.55 a month, inside Apify's $5 monthly free credit.
Next steps
If Apify is not connected to Cowork yet, start with the Cowork + Apify connector setup, then test the prompt above once as a normal Cowork task before you schedule it. For deeper control over Indeed's filters and fields, see the Indeed scraping guide. Want to check the pay before you apply? Benchmark the salary from live job ads. More job-seeker prompts are in Claude prompts that use live web data.