May's HN "Who Is Hiring" Wants Cursor Fluency. LinkedIn Returns 4 People.

The May 2026 "Ask HN: Who is hiring?" thread (item 47975571, posted May 1) crossed a line that most LinkedIn-first recruiters haven't noticed yet. Multiple top-voted posts now list Cursor, Lovable, and Replit as ranked hard requirements, not nice-to-haves. And the candidates who actually fit are nearly invisible to Boolean.

If you're sourcing engineers off LinkedIn skill tags in June 2026, you are sourcing for a 2023 job description. The market moved.

What the May 2026 thread actually says

Read the top postings in order and a pattern jumps out. Tiger Tracks, a performance marketing agency (Inc. 5000 #123, ex-Google founders), posted a Founding AI Engineer contract role asking for "top 1% with Lovable, Replit, or Cursor" and the ability to ship a complex tool (multiple data sources, APIs, server hosting, LLM integration) in under a day. Their interview loop is itself a tool-fluency test: a 30-minute intro, a 45-minute founder conversation, and a 60-minute live build in Lovable or Replit where they watch you ship.

We The Flywheel, a media network out of Ho Chi Minh City, posted an Entry-Level Agentic Engineer role with no formal coding background required. They explicitly care more about "curiosity, initiative, and how you use tools like Claude Code, Cursor, or similar to turn ideas into working output."

This is not an HN quirk. ZipRecruiter JDs now use the same language verbatim: "AI-native development workflow experience, must be fluent with tools such as Cursor, Claude Code, v0.dev, Lovable, or comparable AI engineering tools." A Denver-metro manufacturer is hiring an "AI Builder" for someone who has figured out how to use Claude, Cursor, Replit, Lovable, and n8n. Indeed listings now use the literal phrase "vibe coding" in JDs. Alinea Health wants someone to "bring 'vibe coding' projects to life, fully supporting modern AI tools (Cursor, Replit Agent, Lovable, Claude) to prototype and deploy applications efficiently."

The tool list is the JD. The seniority label is decoration.

The Boolean problem, in one number

Here is what your sourcing pipeline is up against.

4

That is not a typo. A keyword scan for profiles mentioning all three tools together in their headline or bio returns four people on the entire planet. Even single-tool searches collapse fast.

160

Of those 160, the top employer bucket is "Stealth Startup" and 16 of the most common titles are some form of Founder. These are the people building with Cursor, not the people you can poach for a Founding AI Engineer contract. The senior ICs Tiger Tracks wants to pay $150,000 in SF or $125,000 in Chicago to "live and breathe AI native tools" are using these tools daily and not bothering to tag them.

Why? Because there is nothing to tag.

LinkedIn's taxonomy can't catch up

LinkedIn discontinued Skill Assessments entirely. The guidance now is to tag skills to credentials, jobs, projects, or education. Translation: "Cursor" and "Lovable" appear, if at all, as freeform self-reported text. They are not structured filterable skills like "Python" or "AWS."

This isn't a temporary lag. It's structural. Cursor and Lovable have no certification programs. There is no "Cursor Certified Developer" badge for LinkedIn to ingest. Replit has a learning platform but no enterprise-recognized credential. The skills tab will probably never carry these tools the way it carries Java.

So you have two compounding problems. The structured skill data doesn't exist, and the freeform text returns four people.

The tool list is the JD. The seniority label is decoration.

Meanwhile Cursor itself is becoming infrastructure. Their code review agent now resolves 78% of issues found by the time the PR is merged. Stripe's CEO has publicly called Cursor his favorite enterprise AI service and said all roughly 40,000 of their engineers are now AI-assisted. The tool is mainstream. The candidate signal for it is not.

The actual skill is velocity, not the tool name

Here is the contrarian read. Tiger Tracks doesn't actually care whether you prefer Lovable or Replit. They care that you can ship a complex tool (multiple data sources, APIs, server hosting, LLM integration) in under a day. The tools are a proxy for shipped velocity. Boolean for "Cursor" misses this because the real signal is portfolio cadence, not a vendor name on a resume.

This is why the GitHub graph matters more than the LinkedIn skill tab right now. Multiple May 2026 HN posts require "a public GitHub or portfolio with shipped projects we can play with." The recruiters winning this market are reading commit histories, deployed Vercel URLs, and Replit project galleries. They are not running Boolean.

This is exactly the gap we built Refolk for. You describe the person in plain English ("show me people who shipped at least three public LLM tools in the last 90 days using Cursor, Lovable, or Replit, based in SF or remote, open to contract") and Refolk reads across GitHub, LinkedIn, and the open web to rank actual builders. The tool fluency comes from the commit log and the deployed project, not from a skill tag that doesn't exist.

The seniority ladder just collapsed

Look at the May 2026 thread again with this lens. Tiger Tracks offers a "path to Head of AI for the right person." Flywheel offers an entry-level role with no formal coding background required. Both are gated on the same tool fluency. The traditional Senior/Staff/Principal Boolean ladder doesn't map to this market at all.

What you actually want to filter on is shipped portfolio over the last 90 days. A 22-year-old who has deployed 14 Lovable apps and an ex-Stripe staff engineer who has shipped two are competing for the same role, and frequently the 22-year-old wins because the staff engineer is still writing PRDs while the kid is on his fourth iteration.

This is also why the "Vibe Coder" job category now exists as a discrete title. ZipRecruiter lists Vibe Coding jobs at $15 to $43 per hour with required stacks of Replit, Claude, ChatGPT, Perplexity, Lovable, Cursor, Figma, Canva, WordPress, Webflow, Zapier/Make, Airtable, and Supabase. These candidates self-identify as builders or operators, not "engineers." Title-based Boolean for "Software Engineer" misses them entirely. So does title-based Boolean for "Founder," because they aren't founders either. They are a new class and LinkedIn doesn't have a slot for them.

Where the candidates actually live

If LinkedIn returns 4, where are the other thousands? Three surfaces, in order of signal density.

GitHub project commits

The strongest signal is a public repo with frequent commits, a deployed URL, and a README that names the stack. Engineers shipping daily with Cursor often have it visible in commit messages ("cursor agent: refactor auth flow") or in their .cursor or .agent.md config files. A grep across recent commit history surfaces real users in a way LinkedIn cannot.

Replit and Lovable galleries

Both platforms have public project pages. The creators link to their other socials. This is a directly browseable candidate pool that no Boolean string touches.

HN profiles and comments

The people answering "Ask HN: Who wants to be hired?" threads with Cursor and Replit in their bio are a self-selected pool of exactly the right candidates. The number of recruiters scraping this is approximately zero.

This cross-surface stitching is the actual job of a 2026 sourcing pipeline. Refolk does it by default, which is why teams running Cursor developers sourcing or Lovable Replit engineers hiring use it instead of throwing more Boolean at the wall. Ask in plain English, get the right people across all three surfaces, ranked.

What to do this week

If you have an open AI-native engineer recruiting req right now, three concrete moves.

First, stop sourcing on tool names as skill tags. They will never populate. Source on shipped artifacts (a deployed URL, a public repo, a Lovable project) and treat tool names as confirmation, not filter.

Second, rewrite your JD to match May 2026 HN language. If your post says "5+ years of professional software engineering experience," you are filtering out the exact pool the market is paying $150K for. Tiger Tracks doesn't ask for years. They ask for "ship a complex tool in under a day."

Third, build a parallel pipeline off GitHub and the open web, not LinkedIn. The 4 people Boolean returns are not your candidates. They are your false floor. The real pool is at least two orders of magnitude larger and lives on surfaces LinkedIn doesn't index. The point of a tool like Refolk is to make the cross-surface query a single sentence instead of a six-tab workflow.

The May 2026 thread is a leading indicator. By the August "Who is hiring?" thread, this language will be in 30% of posts. By November it will be in the majority. The recruiters who quietly built the off-LinkedIn pipeline in June will be the ones filling reqs in Q4. The ones still tuning Boolean strings will be explaining to their hiring managers why the candidates don't exist.

They exist. They just don't exist on LinkedIn.

FAQ

Why don't AI-native engineers tag Cursor or Lovable as a LinkedIn skill?

Two reasons. LinkedIn discontinued structured Skill Assessments and now ties skills to credentials, jobs, projects, or education, and Cursor and Lovable have no certification programs to anchor against. The result is freeform text only. The second reason is cultural. The engineers using these tools daily treat them like a keyboard or a terminal, not like a discrete skill worth listing. You wouldn't list "VS Code" on your profile either.

Is the "Vibe Coder" category a real role or marketing?

Real, and growing fast. ZipRecruiter and Indeed both list Vibe Coding roles with hourly bands of $15 to $43 and explicit stacks naming Replit, Cursor, Lovable, Claude, Zapier, Airtable, and Supabase. Alinea Health is hiring one. These are not engineering roles in the traditional sense. They are operator-builder roles, and they pay because the work ships. Title-based Boolean for "Software Engineer" misses them entirely.

What's the highest-signal place to source Cursor and Replit users today?

GitHub commit history, public Replit and Lovable project galleries, and HN "Who wants to be hired?" threads, in roughly that order. Look for shipped, deployed projects with recent commits and a stack that includes any LLM API. The candidates with three or more such projects in the last 90 days are the exact pool Tiger Tracks and similar JDs are paying for.

Will LinkedIn's skill taxonomy catch up to these tools eventually?

Probably not in a way that helps sourcing. Without certification programs from Cursor, Lovable, or Replit, there is no enterprise-grade credential for LinkedIn to ingest as a structured skill. The platform will keep relying on freeform text, which means boolean search AI tools coverage stays thin. Expect this gap to persist through 2027 at minimum. Plan your pipeline accordingly.