Job Search With Claude: Find Jobs and Hiring Managers (2026)
By Kushal Magar · October 17, 2026 · 14 min read
Key Takeaway
A job search Claude can run end to end has five steps: pull live LinkedIn jobs with search_linkedin_job_openings, score them against your resume, find the likely hiring manager with find_people, get their work email with find_work_email and confirm it with verify_email, then draft a short tailored note. A full run for 25 jobs and five contacts costs roughly 16 SyncGTM credits. Claude researches and drafts. You apply and send.
A job search Claudecan actually run needs live data, not just a good resume prompt. Connect SyncGTM's MCP and Claude can pull current LinkedIn listings, rank them against your background, find the person who likely owns the opening, and verify their work email before you write a word.
This is the pillar guide of a five-part series. It walks through the full loop with the exact prompt, a sample output and the credit cost for every step, and it is blunt about what Claude should never do alone: auto-apply or mass email.
How do you run a job search with Claude?
Connect https://api.syncgtm.com/mcp to Claude, then prompt it to call search_linkedin_job_openings for live listings, find_people for the likely hiring manager, find_work_email and verify_email for a deliverable address, and to draft a short note for you to review. Claude does the research. You make every send.
TL;DR
- The loop: jobs, resume score, hiring manager, verified email, tailored note.
- Tools:
search_linkedin_job_openings,find_people,find_work_email,verify_email, plus optionalenrich_organizationandjob_openings_growth_rate. - Cost: 0.3 credits per job, 0.2 per person, 1 per work email, 0.1 per verification. About 16 credits for 25 jobs and five contacts.
- Setup: one custom connector or one terminal command. No API key.
- Hard rule: Claude drafts, you send. No auto-apply, no mass email.
Why Run a Job Search Like Outbound
Most guides to using Claude for job hunting stop at resume rewrites, cover letters and interview drills. Those help, but they all begin after you have found a role and end with an application that joins a pile.
Sales teams solved this problem years ago. They pick a target account, find the person who owns the problem, confirm a way to reach them, and send one relevant message. A job seeker can do the same thing: the open role is the signal, the hiring manager is the prospect, and your resume is the proof point.
That is the same pipeline described in our guides to lead generation with Claude and enrichment in Claude, pointed at your own career. If you only want the listings part, the narrower tutorial on job listings in Claude covers it.
Connect SyncGTM MCP to Claude
The Model Context Protocol lets Claude call outside tools. SyncGTM's server lives at one URL and signs in through your browser, so there is no API key to copy.
Option A: Claude app (desktop or web)
- Open Settings, then Connectors, and choose Add custom connector. Custom connectors need a paid Claude plan. On Team and Enterprise plans, a workspace admin has to add it (see Anthropic's custom connector guide).
- Enter the URL
https://api.syncgtm.com/mcpand leave the other fields empty. - Click Authorize, sign in to SyncGTM in the browser and approve. If you belong to more than one SyncGTM organization, pick the right one in the dropdown, because that organization's credits are the ones spent.
- Enable SyncGTM from the tools menu in a new chat.
Option B: Claude Code
Terminal
claude mcp add --transport http syncgtm https://api.syncgtm.com/mcp
Restart Claude Code, run /mcp, select syncgtm and authenticate in the browser. The full walkthrough lives in the SyncGTM MCP setup docs.
Test it with a free call
Prompt
Check my SyncGTM credit balance.
check_credits is free. If it returns a balance, you are connected. New to the setup? The LinkedIn scraper MCP guide covers the same connection in more detail.
The Job Search Claude Workflow in Five Steps
Run the steps in one conversation so Claude keeps the job list, your resume and each shortlisted person in context. The sample tables below are made-up examples that show the shape of the output.
Step 1: Pull live LinkedIn jobs
search_linkedin_job_openings searches LinkedIn listings across all companies. Only title is required (an array, so one call can cover several titles). Everything else narrows the result.
Prompt
Search LinkedIn jobs for "Revenue Operations Manager" and "RevOps Manager" in the United Kingdom. Remote or hybrid, mid-senior level, posted in the past week. Exclude recruiting agencies and drop duplicates. Return 25 results as a table: title, company, location, posted date, job URL.
Behind the scenes, Claude maps that to these parameters:
| You said | Parameter | Value |
|---|---|---|
| Titles | title | ["Revenue Operations Manager", "RevOps Manager"] |
| Where | locations | ["United Kingdom"] |
| Remote or hybrid | work_type | ["remote", "hybrid"] |
| Mid-senior | experience_level | ["mid-senior"] |
| Past week | published_at | "r604800" (r86400 = 24h, r2592000 = month) |
| No agencies | exclude_recruiting_agencies | true |
| Drop duplicates | save_only_unique_items | true |
| 25 results | max_results | 25 (1 to 100, default 10) |
Sample output (first three rows)
| Title | Company | Location | Posted | URL |
|---|---|---|---|---|
| Revenue Operations Manager | Northwind Analytics | London (hybrid) | 2 days ago | linkedin.com/jobs/view/… |
| RevOps Manager | Lumen Freight | Remote, UK | 3 days ago | linkedin.com/jobs/view/… |
| Revenue Operations Lead | Harbor Payments | Manchester (hybrid) | 5 days ago | linkedin.com/jobs/view/… |
Illustrative example with placeholder companies and people. Your results will differ.
Cost: 0.3 credits per job returned, so 25 jobs is 7.5 credits. A search that returns fewer results costs less. For more filters, including salary, applicant count and company size, see the tool reference. If a search comes back empty, drop the stricter quality filters first.
Step 2: Score each role against your resume
You can score in two layers. The tool accepts resume_keywords, an array of keyword and alias pairs, and scores how well each job matches. Then Claude reads each description against your full resume for the judgement a keyword count cannot make.
Prompt
Here is my resume: [paste resume text, remove phone and home address]. Re-run the search with resume_keywords for Salesforce (alias SFDC), HubSpot, SQL, forecasting and territory planning. Then rank the 25 roles from best to worst fit. For each, give a score out of 10, two reasons it fits, one gap, and whether the posting names a salary. Do not claim I have skills that are not on my resume.
Sample ranking
| Rank | Company | Fit | Why it fits | Gap |
|---|---|---|---|---|
| 1 | Northwind Analytics | 9/10 | Salesforce, forecasting, scope matches your last role | Wants dbt experience |
| 2 | Harbor Payments | 8/10 | SQL plus territory planning called out | Fintech background preferred |
| 3 | Lumen Freight | 7/10 | HubSpot to Salesforce migration is your headline win | Needs people-management history |
Illustrative example with placeholder companies and people. Your results will differ.
Cost: no extra call. The scoring rides on the same search, so re-running it costs another 0.3 credits per job. Ask Claude to rank from the results already in the chat instead, and keep the top five.
Optional: before investing in a company, ask Claude to call job_openings_growth_rate (3 credits) or head_count_growth_rate(2 credits) with the company's LinkedIn page. Rising openings and headcount suggest the role is funded, not a stale repost.
Step 3: Find the hiring manager
Job posts rarely name the manager. So you infer the title: someone a level or two above the role, in the same function. For a Revenue Operations Manager, that is usually a Director or VP of Revenue Operations, or a CRO at a small company.
Prompt
For my top 5 roles, find the most likely hiring manager at each company. Use find_people with the company name, job titles like "Director of Revenue Operations", "VP Revenue Operations" and "Head of RevOps", seniority Director or VP, limit 3 per company. Return name, title and LinkedIn URL, and say which one you think owns the role and why.
Sample output for one company
| Name | Title | Why likely |
|---|---|---|
| Jordan Example | VP Revenue Operations | Role sits in RevOps, VP level, based in the same region |
| Sam Placeholder | Director, Sales Operations | Adjacent team, possible stakeholder |
| Alex Sample | Chief Revenue Officer | Owner if RevOps reports to revenue |
Illustrative example with placeholder companies and people. Your results will differ.
find_people takes job_titles, job_levels (C-Team, VP, Director, Manager, Staff, Other), company_names or company_linkedin_urls, and a limit of 1 to 50. Prefer the LinkedIn company URL when you have it, because a name can match the wrong company.
Cost:0.2 credits per person returned. Three people at five companies is 3 credits. This is an inference, not a fact. Check the posting for a named recruiter or hiring team and confirm the person's current role on their profile before you write.
Step 4: Get and verify their work email
find_work_emailtries several data providers in turn. Give it the person's linkedin_url, plus the company domain when you know it (it unlocks extra providers). A name with no domain or LinkedIn URL will not resolve.
Prompt
For the chosen manager at each of my top 5 companies, call find_work_email with their LinkedIn URL and the company domain. Then run verify_email on every address found. Show a table: name, company, email, verification result. Do not keep any address that is not confirmed deliverable.
Sample output
| Person | Company | Verification | |
|---|---|---|---|
| Jordan Example | Northwind Analytics | j.example@northwind-analytics.example | Deliverable |
| Sam Placeholder | Harbor Payments | Not found | Skipped |
| Alex Sample | Lumen Freight | a.sample@lumen-freight.example | Risky, do not use |
Illustrative example with placeholder companies and people. Your results will differ.
Cost: 1 credit per find_work_email call, and misses are not billed. verify_email is 0.1 credits. No verified address? Skip the cold email, send a short LinkedIn connection note yourself, or apply and follow up later. Our email finder MCP guide explains how the waterfall works.
Step 5: Draft a tailored note
The note is where most AI-assisted outreach fails. Give Claude hard constraints so it writes like a person, and keep it grounded in two things only: the posting and your actual resume.
Prompt
Draft a note to each verified contact. Max 90 words. Plain text, no buzzwords. Structure: one line naming the exact role and one specific thing from the job post, one line with a result from my resume (use only facts I pasted), one low-pressure ask. Subject under 7 words. Do not send anything. Save the drafts as a list I can review.
Sample draft (illustrative)
Subject: RevOps Manager role at Northwind
Hi Jordan, I saw the RevOps Manager opening and your focus on forecast accuracy. At my last company I rebuilt the pipeline forecast in Salesforce and cut the monthly variance from 18% to 7%. I have applied through the posting. If it is useful, I can send a two-page summary of how I did it.
Cost:free. Drafting is Claude's own work. Read each draft, check every claim against your resume, edit the wording until it sounds like you, and send it from your own email account. For more outreach structure, see our Claude cold email guide.
What a Full Run Costs in Credits
Credit rates are per call, so you can estimate a run before you start. Here is one pass: 25 jobs, five shortlisted companies, three people each, one email attempt per company.
| Step | Tool | Rate | Volume | Credits |
|---|---|---|---|---|
| Pull jobs | search_linkedin_job_openings | 0.3 per job | 25 jobs | 7.5 |
| Hiring managers | find_people | 0.2 per person | 15 people | 3.0 |
| Work emails | find_work_email | 1 per call (misses free) | 5 calls | up to 5.0 |
| Verification | verify_email | 0.1 per call | 5 calls | 0.5 |
| Total | about 16 |
Rates can change. Confirm the current numbers in the credits estimator and see plan options on the pricing page. Claude can also run check_credits (free) before a big batch.
What Claude Should Not Do Alone
Claude can compress a week of research into an hour. It should not make the decisions that put your name on something. Keep these rules in the prompt and in your habits.
- Do not auto-apply.Bulk-submitted applications carry the wrong details, ignore screening questions and can break a job board's rules. Apply to a short list yourself, with a resume you checked.
- Do not mass email. This workflow finds one person per role. Sending the same note to hundreds of managers is spam, burns your reputation and can breach email laws such as CAN-SPAM and GDPR. A good cap is a handful of personal notes a day.
- Review every message before it leaves. Claude drafts, you read, edit and send from your own account. Never give it the ability to send on your behalf unattended.
- Never let it invent experience. Tell it to use only facts you pasted, and check every number. Fabricated claims surface fast in interviews and background checks.
- Respect LinkedIn's rules. SyncGTM reads public data through a hosted API and never logs in as you. That is different from running bots on your own account. Do not automate connection requests or messages on your profile, and read LinkedIn's User Agreement so you know the terms you operate under.
- Treat the hiring manager as a guess. The lookup ranks likely owners. Confirm before you address someone as the person who hires.
- Protect your data. Remove your phone number and home address before pasting a resume into any chat.
Common Mistakes
- Searching too broadly. Ask for 100 results on a loose title and you pay for noise. Start with 25 and tighten filters.
- Stacking every quality filter. Salary, under-10-applicants and named-recruiter filters each exclude most roles. Add one at a time.
- Skipping verification. A bounced email to a hiring manager looks careless. Verify at 0.1 credits.
- Writing one template. If the note could be sent to any company, it will be ignored. Name the role and one detail from the posting.
- Chasing stale posts. Use a posted-after window and check hiring momentum before you spend credits on contacts.
Go Deeper: The Job Search Claude Series
This post is the overview. Four follow-ups each go deep on a different layer of the same workflow:
- Job Search Claude Skill: build a SKILL.md job hunter that packages these steps into a reusable command.
- Job Search Claude Prompts: 15 copy-paste templates for searching, scoring and drafting.
- Job Search Claude Agent: automate the daily hunt with a scheduled agent, keeping a human in the send loop.
- Job Search Claude Plugin: bundle skills, agents and MCP into one install.
Hiring for revenue and GTM roles specifically? Our map of GTM engineer jobs shows where that demand is growing.
Conclusion
A good job search Claude workflow replaces scrolling with a pipeline: live jobs in, ranked shortlist out, a likely manager and a verified email attached, a draft waiting for your edit. The data steps are cheap, about 16 credits for a serious pass, and the expensive part stays yours: judgement and the send button.
Connect the server, run one search for the role you want most, and walk a single company through all five steps before you scale up. When that note feels like something you would be happy to receive, repeat it for the next four. You can start with SyncGTM for free.
