In May I did one thing on Mercor. I did not add a job, a degree, or a skill I do not have. I rewrote the profile I already had, using a prompt, and an instant offer arrived. I wrote that up at the time in I updated my resume and got an instant Mercor offer.
Since then the prompt has gone out by DM to every paid subscriber who asked, and a lot asked.
This is the first post in a new series. AI tools for nomads: prompts and routines I actually use to get and keep remote AI work. The tool is not the point. The point is that the people hiring for this work are models, and if you are not using a model to talk to them, you are bringing a resume to a vector search.
The part of Mercor everyone skips
People polish the resume PDF. They sweat the AI interview. Then they leave the Profile tab half-filled with whatever the resume parser guessed on day one, and never open it again.
That is the wrong place to be lazy, because of what the profile is for.
Mercor does not read your profile the way a recruiter reads a resume. From what I can see working the platform from both the contributor and the reviewer side (this is my read of the mechanics, not Mercor’s published documentation), it parses the document into a structured representation: skills, domain, seniority, tooling, project shape, rate band, availability. It parses each job the same way. Then it matches by similarity. High similarity surfaces you. Very high similarity can produce an offer with no human in the loop. That is what an instant offer is.
So your reader is an LLM. Write for it.
Three things I found on the live profile page this week
I went through my own Mercor profile tab by tab on September 13 and checked every field against how the prompt was telling readers to fill it. The v1 prompt got three things wrong, and all three are worth knowing even if you never run the prompt.
The Summary field takes about 15 words. A paid subscriber found this the hard way: the field cuts off well before a normal profile summary ends, so the v1 prompt was producing summaries that could not be pasted. This limit is not in Mercor's published docs. It is a firsthand product observation, so test it in your own account. Either way, 15 words is the right target, because the matcher weighs the first nouns heaviest.
Skills are tags, entered one at a time, and the parser's guesses stay until you delete them. My own profile had "Smart Parking Solutions" sitting in Skills. That is a product I once worked on. It is not a term any job description on Mercor contains. It was pure noise in what is probably the highest-weight field on the page, and it had been there for months because I never looked.
Mercor's own LinkedIn widget says verified profiles have 6x higher chances of an offer. That is Mercor's claim, in Mercor's words, on Mercor's page. Roughly half the people who DM me about quiet queues have never clicked it.
What is below the paywall
The rest of this post is the v2 prompt in full, rewritten this week so its output follows Mercor’s profile page section by section (Summary, Education, Work experience, Projects, Certifications, Awards, Coding Profiles, Links, Skills, Languages, Hobbies), in the order you fill it. Around it: the four-step routine I run every time I change target roles, what the output looks like when it is right, the answer to the “does re-targeting flag my account” question, and a reply-to-get-it offer where I fill the template for your role myself.
The four-step routine
Step 1. Pick one role, and copy the whole job description. I know you apply to several tracks at once. So do I. But the profile can only argue one identity at a time, and a profile that argues three matches none of them. Choose the role you want most. Open its page on Mercor and copy the full description text into the prompt. Not the link. Your AI tool cannot open Mercor pages behind your login, and a link gives it nothing to match against.
