Pull open roles off any careers page, with no job board subscription.
hyreflow Native is built by hyreflow and is one of the 54 tools hyreflow gives Claude, ChatGPT and other AI assistants. Through hyreflow, an AI agent can run 20 actions, all of which look something up. Paid hyreflow Native actions cost 0.05 to 0.2 credits per result or per request, from your hyreflow credit balance, and there is no separate account to set up.
hyreflow Native is built by hyreflow. Our own scrapers read open roles straight from company career pages, LinkedIn, Indeed and the German federal board. No vendor account and no key, because we built it. Launching a scrape is free; you pay per role that comes back.
A sourcing and signal front end. Roles come back, and every one of them names a company worth contacting.
With hyreflow connected, you ask for the outcome in plain language and the assistant runs hyreflow Native for you. For example:
You connect hyreflow once over MCP, and hyreflow Native comes with it, next to the other 53 tools. You do not need to run a separate hyreflow Native MCP server.
In Claude, ChatGPT or any other MCP client, add a custom connector named hyreflow with the server URL https://recruit.hyreflow.ai/api/v2/mcp. The MCP setup guide on hyreflow.ai has the exact steps for each app.
A browser window opens. Sign in, choose the workspace the assistant may act on, and approve. The connection is live.
hyreflow Native is part of hyreflow, so it is available as soon as hyreflow is connected.
Tell Claude or ChatGPT what you want done with hyreflow Native in plain language. The assistant picks the hyreflow Native action that fits and runs it through hyreflow.
Using Cursor, Claude Code or another client? Every one is covered in the MCP setup guide.
In a hyreflow run, hyreflow Native is most often paired with hyreflow Agent, Apollo, TheirStack and JobAdder.
pull the real requirements out of every scraped ad
a job ad names a company, this names the person
a broader posting index when you have no company list yet
file the scraped roles as live jobs on the desk
Through hyreflow, an AI agent can run 20 hyreflow Native actions, all of which look something up. Paid hyreflow Native actions cost 0.05 to 0.2 credits per result or per request, from your hyreflow credit balance, and there is no separate account to set up.
| hyreflow Native action | What it does | Cost through hyreflow |
|---|---|---|
| Scrape Arbeitsagentur jobs | Looks something up | free |
| Scrape career pages | Looks something up | free |
| Scrape Indeed jobs | Looks something up | free |
| Scrape LinkedIn jobs | Looks something up | free |
| Scrape StepStone jobs | Looks something up | free |
| Search posts | Looks something up | 0.2 credits per request |
| Places within radius | Looks something up | free |
| Repost scan | Looks something up | 0.8 credits per history page |
| Verify emails | Looks something up | 0.05 credits per result |
| Get Arbeitsagentur jobs | Looks something up | up to 0.15 credits per result |
| hyreflow Native action | What it does | Cost through hyreflow |
|---|---|---|
| Get career pages | Looks something up | 0.05 credits per result |
| Get company posts | Looks something up | 0.2 credits per request |
| Get Indeed jobs | Looks something up | 0.05 credits per result |
| Get LinkedIn jobs | Looks something up | 0.05 credits per result |
| Get LinkedIn profile | Looks something up | 0.2 credits per request |
| Get post reactions | Looks something up | 0.2 credits per request |
| Get profile posts | Looks something up | 0.2 credits per request |
| Get run | Looks something up | free |
| Get StepStone jobs | Looks something up | 0.05 credits per result |
| Get verified emails | Looks something up | free |
One credit is priced on the pricing page. The full price list for every tool is in the pricing tables.
65 hyreflow skills call hyreflow Native as part of the run. A skill is a whole pipeline your agent already knows, end to end.
use whenyou have years of placements in your database and no habit of asking those people who else is good
use whena role has to be filled fast and you want to start with the people who have raised a hand, without the open-to-work filter in LinkedIn Recruiter
use wheneveryone who already holds the title is out of budget or not moving, and the realistic hire is someone one level below
use whenyour team knows half the people in your niche between them, and nobody can see who knows whom
use whenyou know exactly which firms to take people from, and which firms must never be touched
use whenyou are sourcing in a German-speaking market and the search returns fewer people than you know are there
hyreflow ships a written playbook for hyreflow Native, and the agent loads it before the first call. Auth, rate limits, what each call costs, which actions need your approval, and the mistakes worth avoiding. It is the difference between an agent that knows the tool and one that guesses at it.
Read the hyreflow Native docs →HOW in lib/hyreflow_native.py.
linkedin_profile; every other native tool runs on its own.A scrape takes a few minutes.
scrape_* → {request_id, status: "QUEUED"}. Free.get_*, passing that request_id → QUEUED → RUNNING → COMPLETED. On
COMPLETED, result.jobs holds the scraped roles. LinkedIn can also end PARTIAL: also
terminal, result.jobs holds what was scraped, result.failed_scrapes counts what wasn't, and
result.errors lists the failures when upstream names them. Poll every ~10s until COMPLETED, PARTIAL or FAILED — re-polls dedup per job, so a tighter loop costs nothing; the terminal poll bills the same
page whenever it lands.career_pages (a company's own careers site) · linkedin_jobs (by title + location, one company's page,
or specific postings via linkedin_job_urls / linkedin_job_ids) · indeed_jobs · arbeitsagentur_jobs
(the German federal employment board) · stepstone_jobs (Germany's largest commercial job board).
linkedin_jobs and indeed_jobs narrow by seniority and sector: experience_levels (internship,
entry level, associate, mid-senior level, director, executive) and included_industries /
excluded_industries (LinkedIn industry names, e.g. Renewable Energy Power Generation), plus
excluded_titles / excluded_companies. Seniority and industry are labels a language model assigns from
each job's detail page, so they can drift between runs; a job it could not label is kept and flagged in
result.warnings. Industry filters need job details on. Filtered-out jobs are not billed. On LinkedIn a
filtered run takes longer, and fewer jobs than rows does not mean the board ran out. Indeed's first
page is about 15 jobs. A second source can top a run up to about 50 per location, best effort: only in the
US, UK, Germany or France, not with job_types, only after a full first page, at most 5 locations per
run, and within a monthly request quota. A filter there thins that pool and cannot page deeper, and fewer
jobs than rows is normal, so don't re-run the same search.
profile_posts (a person, from linkedin.com/in/<username>) · company_posts (a company page, from
linkedin.com/company/<username>). One call returns up to 50 posts newest-first with full text, engagement
counts, the post URL, when it was posted and any companies tagged. Page back with the next_start each
response returns.
search_posts (alias post_search) finds public posts by what they say: keyword (required), scoped to
date_posted — past-week by default, or past-24h / past-month / any. Narrow to authors whose
headline matches with author_title ("Recruiter", "Talent Acquisition") — the way to catch
recruiting-agency staff announcing a search without knowing their handles. Results are the same post
record a feed returns, newest first (sort_by: "relevance" to rank instead), 10 per page; page forward
with the next_page each response returns (null once the window is exhausted).
post_reactions (get_post_reactions) reads the people who reacted to a LinkedIn post — the engagement
list a feed row's totalReactionCount only counts. Pass the post's postUrl; one call returns up to 10
reactors with name, headline, profile URL and the reaction they left. reaction_type (LIKE, EMPATHY,
PRAISE, INTEREST, FUNNY, APPRECIATION) narrows to one kind; total/total_pages report the whole
set, so page forward with next_page (null at the end).
linkedin_profileget_linkedin_profile reads one person's LinkedIn profile from its URL: name, headline, about, location,
open-to-work and hiring flags, experience newest first (company, title, start/end as YYYY-MM,
is_current), education, skills, certifications, languages. No email or phone. It runs second in
the linkedin_profile waterfall, so reach it through linkedin_profile for a list; a URL with no profile
behind it returns success: false and is still one charged request.
places_within_radius (alias postcode_radius) turns a full postcode/ZIP or a city name (place) and a radius (unit mi or
km, up to 25 km; country ISO-2) into the towns, villages and suburbs inside it, nearest first (50 by
default, limit up to 200), as a
locations list for people_search. People databases have no radius filter, so this town list is how a
"within 3 miles of a postcode" or "within 20 km of a city" search runs.
verify_emails checks each inbox for 1 to 50,000 addresses: launch (charged per address), poll get_verified_emails (free)
until COMPLETED (or FAILED), then read result.emails — only the addresses safe to send to. Every other address
is withheld and counted in result.withheld. Email enrichment and personal email enrichment already run this check on every address
they find. Ask for Enrichley, LeadMagic or Icypeas by name to use one of them instead.
repost_scan reads a company's PredictLeads job-opening history for the lookback (domain, plus its
parent or subsidiary domains as related_domains) and returns only the roles that show a signal, each
with its listings as evidence: repost (a fresh listing after an earlier one of the same role at the same
site ran 21+ days and stopped), long-open (live 90+ days) or evergreen (three or more monthly relists,
the latest live). titles or onet_codes narrow it to one role. Renamed roles come back in
needs_review for a read of the two descriptions; warnings flags a cut-off history or stale crawl data.
Job scrapes: 0.05 credits per job returned, charged on retrieval — not per call (Arbeitsagentur adds
+0.1 credits per job with recruiter details, up to 0.15 credits per job). You are charged for the
jobs a poll actually delivers, so size the request to the ask: rows on the launch bounds what the scraper
collects (Indeed defaults to 25), limit on the poll bounds what one page returns (10), and advancing to a
further page bills that page's jobs. Post feeds, post search and post reactions: 0.2 credits
per request (a feed page is up to 50 posts, a search or reactions page up to 10), charged whether or not the handle
resolves or the keyword matches. Profile lookup: 0.2 credits per request, found or not (free on a miss
inside the linkedin_profile waterfall). Postcode radius: free. Repost scan: 0.8 credits/page for each PredictLeads history page it pulls
(one page per domain holding up to 1,000 listings, up to three pages for a domain whose history runs
past that); the scan itself adds nothing.
Email verification: 0.05 credits per address submitted, charged at launch; polling is free.
The sourcing front end for a hiring-signal pipeline: scrape open roles → identify the hiring company → enrich the decision-maker → sequence. Post feeds add the timing layer — read what an account has been posting to open on a launch, a funding round or a hiring push rather than on nothing. Post search finds the accounts in the first place — who said "we're hiring" this week — before a feed reads one of them. Email verification closes the enrich step: verify the found addresses before they reach a sequence.
Yes. hyreflow is a connector for Claude, ChatGPT and other AI assistants. Once hyreflow is connected, you ask for the outcome in plain language and the assistant calls hyreflow Native for you. hyreflow exposes 20 hyreflow Native actions this way.
No. hyreflow Native is built by hyreflow, so there is nothing to connect beyond hyreflow itself.
Paid hyreflow Native actions cost 0.05 to 0.2 credits per result or per request, from your hyreflow credit balance, and there is no separate account to set up. 8 actions are free. hyreflow can show the cost of a paid action before it runs.
Add the hyreflow MCP server to your assistant and sign in. hyreflow Native is part of hyreflow, so it is available as soon as hyreflow is connected.
Connect hyreflow to Claude or ChatGPT and ask for the outcome. It reads the playbook above before it makes a single call.